Skip to main content

Full text of "Differential Psychology Individual"

See other formats

15?    AJ5d  62-05918 


Differential  psychology 



NEW    YORK    •    BOSTON    '    CHK'AGO    •    DALLAS 

MACMILLAN  AND    CO,,   LiMtTia* 

LONDON    •    BOMBAY    •    CAUTTTA    *    MAl>KAh 




Individual  and  Group  Differences 
in  Behavior 


Chairman^  Department  of  Psychology , 
tytcens  College,  New  York 


COPYRIGHT,   1937, 

ALL     RIGHTS     RESERVED — NO     PART     OF     THIS     BOOK     MAY     BE 


Published  May,  1937.    Reprinted  January,  1:939 ; 
May,  1940. 

SET    OT    AND    ELECTROTYPED    BY    T.    MOREY    &    SON 


The  present  book  was  written  with  three  major  aims 
in  mind.  In  the  first  place,  differential  psychology  is 
here  presented,  not  as  a  separate  field  of  psychology,  but 
as  one  approach  to  the  understanding  of  behavior.  Its 
fundamental  problems  are  no  different  from  those  of 
general  psychology.  It  is  apparent  that  if  we  can  explain 
satisfactorily  why  individuals  react  differently  from  each 
other,  we  shall  understand  why  each  individual  reacts  as 
he  does.  The  data  of  differential  psychology  should  help 
to  clarify  the  fundamental  mechanisms  of  behavior  in 
general.  It  is  from  this  point  of  view  that  the  problems 
of  individual  and  group  differences  are  surveyed  in  the 
present  text. 

A  second  aim  of  the  book  has  been  to  coordinate  the 
various  topics  which  have  heretofore  been  loosely  joined 
together  under  the  caption  of  "individual  differences." 
With  the  rapid  increase  in  material  on  any  one  of  these 
topics,  there  has  appeared  a  tendency  to  regard  each 
segment  as  an  independent  field  of  investigation.  The 
mutual  interrelation  of  the  various  problems  has  become 
obscured  by  the  accumulation  of  data  at  a  more  rapid 
rate  than  it  could  comfortably  be  assimilated.  It  is  an 
undoubted  fact  that  the  various  branches  of  differential 
psychology  have  undergone  a  phenomenal  development 
during  the  past  two  decades.  This  has  resulted  in  an 
increasing  specialization  of  interest  among  research  work- 
ers and  a  frequent  disregard  of  the  broader  implications 
of  the  data.  Many  apparently  unrelated  aspects  of  differ- 
ential psychology  can  profitably  be  considered  together. 


Thus  the  data  on  such  questions  as  familial  resemblance, 
the  relationship  between  mental  and  physical  traits,  the 
growth  of  abilities,  and  the  organization  of  mental  traits 
can  be  shown  to  have  a  mutual  bearing  upon  each  other. 
The  writer  has  endeavored  to  bear  constantly  in  mind  the 
interrelationships  among  different  types  of  investigations 
and  has  attempted  to  present  a  systematic  organization 
and  integration  of  the  material. 

Finally,  it  has  been  our  aim  to  report  the  major  data 
and  conclusions  of  differential  psychology  in  a  form  suit- 
able for  the  college  student.  The  present  book  has,  in 
fact,  grown  out  of  a  course  in  differential  psychology 
which  the  writer  has  taught  at  Barnard  College  since 
1930.  Much  of  the  material  of  differential  psychology  has 
heretofore  been  available  only  in  journal  articles  and 
highly  technical  books.  Consequently,  certain  topics  have 
been  customarily  omitted  or  touched  upon  lightly  in  dis- 
cussions of  individual  differences,  being  considered  too 
" difficult"  for  the  elementary  student.  Mental  organ- 
ization is  a  good  example  of  such  a  topic,  although  by 
no  means  the  only  one.  Even  the  more  advanced  student 
of  psychology  who  has  specialized  in  other  phases  of  the 
subject  often  finds  it  impossible  to  keep  informed  on 
current  developments  in  certain  branches  of  differential 
psychology.  The  writer  believes  that  a  non-technical  and 
easily  comprehensible  presentation  of  such  material  is 
both  feasible  and  desirable.  An  understanding  of  the 
basic  concepts  and  major  findings  of  any  problem  need 
not  be  limited  to  those  who  have  mastered  its  specialized 

The  present  book  could  serve  either  as  a  text  in  special 
courses  on  differential  psychology  and  "individual  differ- 
ences" or  as  a  supplementary  text  in  such  courses  as 
general,  applied,  or  educational  psychology,  and  mental 



testing.  It  also  provides  relevant  material  for  courses  in 
social  psychology. 

There  is  at  present  a  growing  interest  in  the  origin  of  psy- 
chological differences  among  individuals  and  especially 
among  groups.  Many  of  the  problems  of  differential  psy- 
chology have  become  the  subject  of  frequent  popular  dis- 
cussion and  controversy.  Thus  the  subject  matter  of  this 
branch  of  psychology  should  prove  intrinsically  valua- 
ble. Moreover,  as  has  already  been  pointed  out,  famili- 
arity with  the  main  principles  of  individual  variation  will 
strengthen  the  student's  understanding  of  general  psy- 
chology. Lastly,  the  analysis  of  the  problems  and  difficulties 
encountered  in  the  investigation  of  individual  differences 
should  foster  a  critical  attitude  towards  all  data  on  these 
questions.  Throughout  the  present  book,  special  emphasis 
has  been  placed  upon  the  examination  of  common  sources  of 
errors  and  pitfalls  in  the  interpretation  of  obtained  facts. 
We  have  thus  hoped  to  provide  the  student  with  certain 
tools  whereby  he  may  evaluate  for  himself  a  set  of  data 
with  which  he  is  confronted.  It  seemed  to  the  writatrthat 
this  was  more  valuable  than  a  mere  compendium  of  facts. 
The  development  of  critical  ability  and  of  a  dispassionate 
and  objective  attitude  towards  human  behavior  is  an 
urgent  need  in  a  world  of  rapidly  changing  values. 

The  .writer  is  indebted  to  Prof.  H.  L.  Hollingworth  of 
Barnard  College  for  a  critical  reading  of  the  manuscript 
and  for  many  stimulating  comments.  She  wishes  to 
acknowledge  the  invaluable  assistance  of  Dr.  J.  P.  Foley, 
Jr.,  of  The  George  Washington  University  for  an  inten- 
sive reading  of  the  entire  manuscript  in  its  early  stages 
and  for  countless  specific  suggestions.  Grateful  acknowl- 
edgment is  also  made  to  the  many  authors  whose  works 
have  been  quoted  in  this  book  and  especially  to  those 
who  have  made  accessible  to  the  writer  unpublished  data 

viii  PREFACE 

from  their  investigations.  Finally,  thanks  are  extended 
to  the  various  publishers  who  have  generously  granted 
permission  for  the  reproduction  of  figures  and  quotations 
from  their  publications.  In  all  of  these  cases,  specific 
acknowledgment  has  been  made  in  the  appropriate  places. 

April,  1937 





Individual  Differences  in  Pre-experimental  Psycho- 
logical Theory 5 

The  Personal  Equation  in  Astronomy 9 

The  Rise  of  Experimental  Psychology       ....  10 

Galton  and  the  Biological  Influence      ......  n 

Early  Experimentation  with  Tests 13 

Beginnings  of  Differential  Psychology 16 

Intelligence  Testing 18 

Group  Testing 21 

The  Measurement  of  Personality 23 

Current  Trends  in  Differential  Psychology     ...  24 


The  Distribution  of  Individual  Differences      ...  30 

The  Normal  Curve 32 

Other  Types  of  Distribution  Curves  and  What  They 

Mean 35 

Some  Typical  Distributions 42 

Extent  of  Variability  in  Different  Traits    ....  51 

Individual  Differences  in  Infrahuman  Groups      .      .  57 


Fundamental  Concepts    .           67 

Prenatal  Environment Ji 

Experimentally  Produced  Variations  in  Behavior.     ,  74 

Human  Children  Reared  in  Abnormal  Environments  77 

Differences  among  Social  or  Occupational  Groups     .  83 




The  Families  of  Eminent  Men 97 

Degenerate  Families 101 

Parent-Child  Resemblance 106 

The  Comparison  of  Siblings no 


The  Psychological  Study  of  Twins 119 

Twin  Resemblances  in  Mental  Traits 121 

Case  Studies  of  Identical  Twins  Reared  Apart     .      .  125 

Foster  Children 129 


Experiments  on  Infant  Development 140 

The  Effects  of  Special  Training  and  Coaching  upon 

Mental  Test  Performance 145 

The  Problem  of  Practice  and  Individual  Differences  149 
Typical  Experimental  Findings  on  Practice  and  Vari- 
ability           .....  156 


The  Growth  Curve 161 

Adult  Intelligence 170 

Adult  Learning 175 

Training  and  Growth       .            1 80 


MENTAL  AND  PHYSICAL  TRAITS  .     .     .  186 

Cranial  Measurements 187 

Facial  Characteristics 191 

Body  Build      , 196 

Physiological  Conditions 203 

Sensory  Defects 210 

General  Evaluation 215 

TIONAL TYPES .     .  221 

Type  Theories 222 

Implications  of  Type  Psychology 230 



Evidence  from  Abnormal  Cases 234 

Correlational  Studies  on  Normal  Groups  ....  242 

Investigations  on  "Pure  Types" 246 

VIDUAL   258 

The  "Idiot  Savant" 260 

Asymmetry    of    Development   in    "Normal"    Indi- 
viduals      265 

The  Individual  Psychograph 269 

The  Measurement  of  Trait  Variability       ....  280 

Evidence  from  Correlation 285 

Major  Contemporary  Theories  of  Mental  Organiza- 
tion     299 

Methodological  Problems 305 

Group  Factors 311 

Form  versus  Content 313 

Shifting  Components  of  Mental  Life 315 

The  Role  of  Environment 318 



Historical  Views  of  Mental  Disorder 329 

The  Concept  of  Abnormality 331 

The  Feebleminded 334 

The  Insane 340 

The  Neurotic 344 

Abnormality  in  Different  Cultures 348 

Abnormality  in  Infrahuman  Organisms     ....  350 

The  Nature  of  Abnormality 351 


Theories  on  the  Nature  of  Genius 357 

Methods  Employed  in  the  Study  of  Genius    .      „     .  362 



Surveys  of  Adult  Genius       ........  365 

The  Early  Mental  Traits  of  Genius 370 

The  Child  Prodigy 373 

Mental  Tests  and  the  Superior  Child 376 

Subsequent  Careers  of  Intellectually  Gifted  Children  '  380 


PROBLEMS .     .     .•  386 

Sex  Differences  in  Achievement 387 

Sex  Differences  in  Variability 390 

Selective  Factors 394 

Relative  Maturity 396 

Reliability  of  a  Group  Difference    ......  399 

Overlapping 402 

Nature  of  the  Measuring  Instrument 404 

Sex  Differences  and  Culture 407 



Sensory  Acuity  and  Motor  Ability 415 

Special  Aptitudes 417 

General  Intelligence 425 

Scholastic  Achievement »  428 

Interests,  Preferences,  and  Attitudes 432 

Character  Traits 436 

Emotional  Adjustment 439 

A  " Masculinity-Femininity"  Index  of  Personality 

Traits 443 

LEMS OF  GROUPING      . 451 

What  Is  a  Race?  ."...." 454 

Evaluation  of  the  Criteria  of  Race        .....  458 

A  Tentative  Classification  of  Racial  Groups   .      .      .  461 

National  and  Linguistic  Groupings 464 

Race  Mixture "463 

Psychological  Studies  of  Hybrid  Groups    ....  468 

Immigrant  Groups  - 475 




Language  Handicap 483 

Difficulties  of  Test  Administration 490 

Educational  Differences 492 

Social  and  Economic  Status 495 

General  Cultural  Milieu 497 

Traditions,  Customs,  and  Interests 503 

The  Criterion  of  Intellectual  Superiority   ....  508 

The  Comparative  Achievements  of  Different  Races  512 



The  Army  Data 518 

"Natio-Racial"  Differences  among  Children  of  Immi- 
grant Groups 523 

Cross-Comparisons  among  Racial  and  National 

Groups  in  Europe 527 

Race  versus  Culture  in  the  Development  of  Person- 
ality Traits  .  . , 53 1 

Gesture:  An  Example  of  Culturally  Conditioned  Be- 
havior    538 

General  Evaluation 543 

TIONS    548 

Theories  of  Urban-Rural  Differences  in  Intelligence  551 
Mental  Test  Surveys  of  Rural  Schoolchildren  .  .  554 
City  and  Country  Differences  in  Europe  ....  557 
The  Intelligence  of  Mountain  Children  ....  560 
The  Specificity  of  Intellectual  Differences  .  .  .  565 
The  Mental  Development  of  Farm  Children  .  .  .  572 
Some  Data  on  Selective  Migration  versus  Environ- 
mental Handicap 574 



Structure  and  Function 581 

Experiential  Determination  of  Behavior    .      ...     .  583 



Specificity  of  Behavioral  Development      ,  588 

"Normal"  and  " Natural"  Behavior    .....  595 

Nature  and  Variety  of  Psychological  Groups  .      ,     .  598 

The  Meaning  of  Individuality 601 




No  topic  has  greater  significance  for  the  organization 
of  life  among  human  beings  than  that  of  the  nature  and 
basis  of  the  individual  differences  among  those  human 
beings.  Except  for  individual  differences  among  us  there 
would  be  no  such  distinctions  as  right  and  wrong;  just 
and  unjust;  health  and  illness.  There  would  be  no  laws, 
no  courts,  no  systems  of  ethics;  no  politics  and  no  need 
of  government.  Individual  differences  are  responsible 
for  such  institutions  as  education,  for  such  episodes  as 
wars,  and  probably,  if  the  truth  were  known,  for  culture, 
for  science,  for  the  church,  and  for  nearly  everything 
else  that  is  characteristically  human. 

Another  striking  fact  is  the  way  in  which  activities 
and  institutions,  springing  from  human  individual  differ- 
ences, become  in  turn  responsible  for  further  differentia- 
tion and  variation.  There  can  be  no  doubt  that  human 
diversity  in  some  characteristics  is  biological  in  origin, 
and  that  to  some  extent  we  are  different  one  from  an- 
other because  of  the  different  stocks  from  which  we  spring 
and  because  of  the  multitudinous  influences  that  make 
variation  intrinsic  to  all  living  things.  It  is  still  true  that 
"Men  do  not  gather  grapes  of  thorns  nor  figs  of  thistles." 

It  is  equally  true  that  much  of  human  diversity,  es- 
pecially in  the  more  subtle  psychological  traits,  is  a  re- 
flection of  the  incidence  of  differential  environmental 
effects.  Theories  of  the  basis  of  individual  differences 
themselves  exhibit  variation.  They  range  from  a  dogmatic 
hereditarianism,  with  a  somewhat  aristocratic  air  of  de- 
terminism, through  sane  and  balanced  recognitions  of 




Man  has  always  been  aware  of  differences  among  his 
fellow-beings.  He  has,  to  be  sure,  entertained  various 
theories,  beliefs,  or  superstitions  regarding  the  causes  of 
such  differences,  and  has  interpreted  them  differently 
according  to  his  own  traditional  background,  but  he  has 
at  all  times  accepted  the  fact  of  their  existence.  Among 
primitive  -peoples,  unusual  deviations  in  behavior  are 
clearly  recognized, ,  Thus  many  primitive  groups  acknowl- 
edge exceptional  -  artistic  talent  among  their  members 
and  encourage  the  development  of  --specialized  artists. 
The  presence  of  hysterical  or  epileptoid  symptoms,  para- 
noid trends,  and  similar  peculiarities  of  behavior  has 
frequently  been  regarded' as  an  index  of  religious  or  magi- 
cal powers  and  has  been  treated  accordingly.  The  history 
of  religion  is  replete  with* such  instances.  At  any  level 
of  cultural  development,  specialization  of  labor  itself  im- 
plies a  tacit  assumption  of  differences -among  people. 

Nor  is  this  response,  to  individual  -differences  limited 
to  the  human  species.  Instances  from  inf  rahuman  behavior 
caritTcadily  be  found.  The  acceptance  of  .-certain.' individut 
ab'-as  "leaders'4 -by  herds -.of  elephants,  buffaloes,  and 
similar .-gxegariotis  •  animals  tes  -been:- widely  ;discussed  in 
the  literature. both  of  feet  and  driiction.:  In, "communities 
of  baboons,  .^fceitain  member  is  posted  .'as  /c sentinel"  to 
watch  for  the  approach  :of.  danger  -and"  warn  the  others 
by  conventional -cries.  "The  frequently- -described  "hack- 
ing" or  hen-pecking  behavior  of  chickens  is  another  case 
in  point*  A  definite -relationship  of  social  domination  is 



often  displayed  by  chickens  In  the  barnyard,  this  fighting 
or  "hacking"  behavior  usually  centering  about  the  ac- 
quisition of  food.  In  such  cases,  A  will  attack  B3  although 
the  reverse  will  not  occur.  Violent  conflicts  often  ensue 
when  the  authority  of  the  chief  "hacker"  in  the  group  is 
disputed.  These  and  many  other  examples  all  serve  to 
suggest  a  widespread  differential  response  to  individuals 
within  a  group. 

The  objective  and  quantitative  investigation  of  indi- 
vidual differences  in  behavior  phenomena  is  the  domain 
of  differential  psychology.  What  is  the  nature  and  extent 
of  such  differences?  What  can  be  discovered  as  to  their 
causes?  How  are  the  differences  affected  by  training, 
growth,  physical  conditions?  In  what  manner  are  the 
differences  in  various  traits  related  to  one  another,  or 
organized?  These  are  some  of  the  fundamental  questions 
raised  by  differential  psychology  and  will  be  treated  in 
Part  I  of  the  present  book. 

Differential  psychology  is  also  concerned  with  an 
analysis  of  the  nature  and  characteristics  of  major  tradi- 
tional groupings,  such  as  the  subnormal  and  the  genius, 
the  sexes,  and  racial,  national,  and  cultural  divisions* 
This  furnishes  the  subject-matter  of  Part  II.  The  study 
of  such  group  differences  serves  a  threefold  purpose.  In 
the  first  place,  one  cannot  ignore  the  fact  that  such  group- 
ings are  being  made  in  the  practical  realm  of  everyday 
life.  These  distinctions  cannot  be  swept  away  casually 
on  the  grounds  that,  perhaps,  the  study  of  individual 
differences  reveals  no  need  for  them  or  -for  any  sharp 
divisions  into  clear-cut  categories.  Certain  groups  are 
recognized  and  responded  to  as  distinctive  in  our  present- 
day  society.  For  a  purely  practical  reason,  therefore, 
these  groups  must  be  investigated,  in  the  hope  that  the 
specific  findings  may  throw  some  light  upon  their  nature 


and  possibly  further  a  more  intelligent  practical  under- 
standing of  them. 

Secondly,  the  comparative  investigation  of  different 
groups  should  help  to  clarify  the  fundamental  problems 
of  individual  differences  in  general.  In  such  groups  we 
can  see  the  principles  of  individual  differences  in  operation 
and  can  note  their  effects.  Group  differences  in  behavior, 
when  considered  in  conjunction  with  other  concomitant 
differences  among  the  groups,  furnish  an  excellent  avail- 
able means  of  analyzing  the  causes  of  variability. 

Thirdly,  the  comparison  of  a  psychological  phenomenon 
as  it  occurs  in  different  groups  may  contribute  towards  a 
clearer  understanding  of  the  phenomenon  itself.  The  find- 
ings of  general  psychology,  when  tested  in  widely  varying 
groups,  are  sometimes  found  to  be  not  so  general  as  was 
supposed.  To  study  a  phenomenon  in  all  its  varied 
manifestations  is  to  have  a  better  grasp  of  its  essential 

Notwithstanding  the  early  and  widespread  recognition 
of  individual  differences  in  the  practical  adjustments  of 
everyday  life,  the  systematic  investigation  of  such  differ- 
ences is  a  relatively  recent  development  in  psychology. 
A  brief  historical  survey  of  theories  as  well  as  investiga- 
tions in  the  field  of  differential  psychology  will  be  given. 


One  of  the  earliest  instances  of  explicit  recognition  of 
individual  differences  is  to  be  found  in  the  Republic  of 
Plato.  A  fundamental  aim  of  Plato's  ideal  state  is,  in 
fact,  the  assignment  of  individuals  to  the  special  tasks 
for  which  they  are  suited.  In  Book  II  of  the  Republic 
appears  the  following  statement:  "Really,  I  said,  it  is 
not  improbable;  for  I  recollect  myself,  after  your  answer, 


that,  In  the- first  place,  no  two  persons  are  born  exactly 
alike,  but  each  differs  from  each  in  natural  endowments, 
one  being  suited  for  one  occupation  and  another  for  an- 
other" (21,  p.  60). 1  Plato  proposes  a  series  of  "actions 
to  perform "  for  use  as  tests  of  military  aptitude  on  those 
who  are  to  be  the  soldiers  of  his  ideal  state.  These  actions 
are  designed  to  sample  the  various  traits  considered  essen- 
tial to  military  prowess,  and  represent  the  first  systematic 
description  of  an  aptitude  test  on  record. 

Nor  did  the  versatile  genius  of  Aristotle  overlook  indi- 
vidual variation.  He  discusses  at  some  length  group 
differences,  including  species,  racial,  social,  and  sex  differ- 
ences in  mental  and  moral  traits.  In  many  of  his  works 
there  is  also  an  implicit  assumption  of  individual  differ- 
ences, although  it  is  interesting  to  note  that  Aristotle 
does  not  offer  any  extensive  treatment  of  these  differences 
as  such.  One  gets  the  impression  that  he  regards  the  exist- 
ence of  individual  variation  as  too  obvious  to  need  special 
mention.  That  he  attributes  such  differences  at  least  in 
part  to  innate  factors  seems  to  be  indicated  by  a  number 
of  statements  such  as  the  following: 

Perhaps,  then,  some  one  may  say,  "Since  it  is  in  my 
power  to  be  just  and  good,  if  I  wish  I  shall  be  the  best  of 
all  men/'  This,  of  course,  is  not  possible.  .  .  .  For  he  who 
wills  to  be  best  will  not  be  so,  unless  Nature  also  be  presup- 
posed (i,  Magna  Moralia,  ir87b). 

Throughout  the  several  Ethics  of  Aristotle,  there  appear 
passages  which  imply  individual  variation-  The  follow- 
ing statement,  for  example,  leaves  little  doubt  regarding 
Aristotle's  position  on  this  point* 

After  these  distinctions  we  must  notice  that  in  everything 
continuous  and  divisible  there  is  excess,  deficiency,  and  the 

1  The  numbers  in  parentheses  refer  to  the  numbered  References  at  the  end 
of  the  chapter. 


mean,  and  these  In  relation  to  one  another  or  in  relation  to 
us,  e.g.,  in  the  gymnastic  or  medical  arts,  and  in  those  of 
building  and  navigation,  and  in  any  sort  of  action,  alike 
scientific  and  non-scientific,  skilled  and  unskilled.  For  mo- 
tion is  continuous,  and  action  is  motion  (i,  Ethica  Eudemia, 


Aristotle  then  proceeds  to  describe  the  characteristics  of 
men  possessing  an  excess  or  a  deficient  amount  of  various 
traits  such  as  irascibility,  audacity,  shamelessness,  and 

In  the  Scholasticism  of  the  Middle  Ages,  individual 
differences  were  completely  neglected.  The  individual 
as  such  had  no  part  in  the  doctrines  and  generalizations 
of  the  Schoolmen.  Rationalistic  methods  persisted  even 
longer  in  the  study  of  mental  life  than  in  other  disciplines, 
where  empiricism  was  gradually  making  itself  felt.  The 
keen  observation  of  the  ancients  which  had  led  to  the 
discovery  of  many  facts  about  human  traits  and  behavior 
was  now  subordinated  to  religious  speculation.  The  logical 
techniques  of  the  Greeks  were  being  employed  primarily 
to  build  elaborate  justifications  for  current  theological 
dogmas.  The  spread  of  Christianity  itself,  with  its  em- 
phasis upon  otherworldliness  and  spiritual  equality,  was 
also  doubtlessly  influential  in  turning  attention  away  from 
the  individual 

The  many  varieties  of  Associationism  which  flourished 
from  the  seventeenth  to  the  nineteenth  centuries  likewise 
took  little  heed  of  individual  differences.  It  was  with  the 
elaborate  mechanics  whereby  ideas  become  associated, 
giving  rise  to  complex  mental  processes,  that  the  associa- 
tionists  were  primarily  concerned.  Their  statements  were 
general  principles  with  no  allowance  for  individual  varia- 
tion. Bain,  the  last  of  the  so-called  pure  associationists, 
does,  however,  give  some  attention  to  individual  differ- 


ences  in  his  writings.  The  following  passage  is  taken 
from  his  book  on  The  Senses  and  the  Intellect  (1855). 
"There  is  a  natural  force  of  adhesiveness,  specific  to 
each  constitution,  and  distinguishing  one  individual  from 
another.  This  property,  like  almost  every  other  assign- 
able property  of  human  nature^  1  consider  to  be  unequally 
distributed"  (2,  p.  237). 

A  simultaneous  development  in  educational  theory 
should  probably  be  included  at  this  point.  In  the  writ- 
ings and  practices  of  a  group  of  "naturalist1"  educators 
of  the  late  eighteenth  and  early  nineteenth  centuries, 
including  Rousseau,  Pestalozzi,  Herbart,  and  Froebel, 
there  is  found  a  definite  shift  of  interest  to  the  individual 
child  (cf.  1 6).  Educational  policies  and  methods  were 
to  be  determined,  not  by  external  criteria,  but  by  direct 
observation  of  the  child  and  his  capacities.  The  emphasis 
still  seemed  to  be,  however,  on  the  observation  of  the 
individual  as  representative  of  individuals  in  general, 
rather  than  as  distinct  from  other  individuals*  Although 
statements  can  be  found  in  the  writings  of  these  educators 
to  the  effect  that  individuals  differ  and  that  their  educa- 
tion should  be  adapted  to  these  differences,  still  the 
emphasis  is  laid  more  heavily  upon  free3  "natural"  edu- 
cation in  contrast  to  externally  and  arbitrarily  imposed 
procedures,  rather  than  upon  individual  differences  them- 
selves. The  term  "individual"  is  often  used  to  mean 
simply  "human  nature." 

Finally,  mention  may  be  made  of  the  various  treatises 
on  race  and  racial  psychology  which  appeared  in  the  late 
eighteenth  and  early  nineteenth  centuries.  Discussions 
of  race  differences  are  to  be  found  in  the  works  of  such 
writers  as  Buffon,  Herder,  and  de  Gobineau,  the  last 
having  been  especially  influential  in  determining  subse- 
quent popular  beliefs  about  race. 


Curiously  enough,  the  first  systematic  measurements 
of  individual  differences  were  undertaken  not  in  psychol- 
ogy but  in  the  old  and  time-honored  science  of  astronomy. 
In  1796,  Maskelyne,  the  astronomer  royal  at  the  Green- 
wich Observatory,  dismissed  Kinnebrook,  his  assistant, 
because  the  latter  observed  the  times  of  stellar  transits 
nearly  a  second  later  than  he  did.  The  method  employed 
at  the  time  to  make  such  observations  was  the  "eye  and 
ear"  method.  This  method  involved  not  only  coordination 
of  visual  and  auditory  impressions,  but  also  rather  com- 
plex spatial  judgments.  The  observer  noted  the  time  to 
a  second  on  the  clock,  then  began  to  count  seconds 
with  the  heard  beats  of  the  clock,  at  the  same  time 
watching  the  star  as  it  crossed  the  field  of  the  telescope. 
He  noted  the  position  of  the  star  at  the  last  beat  of  the 
clock  just  before  it  reached  the  " critical"  line  in  the  field; 
then,  similarly,  he  noted  its  position  with  the  first  beat 
immediately  after  it  had  crossed  that  line.  From  these 
observations,  an  estimate  was  made  in  tenths  of  a  second 
of  the  exact  time  when  the  star  crossed  the  critical  line. 
This  was  the  accepted  procedure  and  was  regarded  as 
accurate  to  one-  or  two-tenths  of  a  second. 

In  1816,  Bessel,  astronomer  at  Konigsberg,  read  of  the 
Kinnebrook  incident  in  a  history  of  the  Greenwich  Astro- 
nomical Observatory,  As  a  result,  he  became  interested 
in  measuring  what  came  to  be  known  as  the  "personal 
equation"  of  different  observers.  Originally,  the  personal 
equation  referred  to  the  difference  in  seconds  between  the 
estimates  of  two  observers.  Bessel  collected  and  published 
data  on  several  trained  observers,  and  pointed  out  not 
only  the  presence  of  such  a  personal  equation  or  error 
when  comparing  any  two  observers,  but  also  the  variabil- 
ity in  the  equation  from -time  to  time.  This  represents 


the  first  published  record  of  quantitative  data  on  indi- 
vidual differences. 

Many  astronomers  followed  up  BessePs  measurements. 
In  the  latter  half  of  the  nineteenth  century,  with  the 
introduction  of  chronographs  and  chronoscopes,  it  became 
possible  to  measure  the  personal  equation  of  a  given 
observer  without  reference  to  any  other  observer.  The 
attempt  was  made  to  reduce  all  observations  to  their 
objectively  correct  values  without  reference  to  a  system 
of  time  based  upon  one  observer  as  a  standard.  Astrono- 
mers also  undertook  an  analysis  of  the  various  conditions 
which  affected  the  size  of  the  personal  equation.  It  was 
this  latter  problem,  rather  than  the  measurement  of  indi- 
vidual differences,  which  was  taken  up  by  the  early  experi- 
mental psychologists  in  their  studies  of  "reaction  time." 


During  the  latter  half  of  the  nineteenth  century,  psy- 
chology began  to  venture  away  from  its  armchair  and 
enter  the  laboratory.  Most  of  the  early  experimental 
psychologists  were  physiologists  whose  experiments  gradu- 
ally came  to  take  on  a  psychological  tinge.  As  a  result, 
both  the  viewpoints  and  the  methods  of  physiology  were 
frequently  carried  over  directly  into  the  infant  science 
of  psychology.  In  1879,  Wilhelm  Wundt  established  the 
first  laboratory  of  experimental  psychology  at  Leipzig. 
Experiments  of  a  psychological  nature  had  been  per- 
formed previously  by  Weber,  Fechner,  Helmholtz,  and 
others,  but  Wundt's  laboratory  was  the  first  to  be  devoted 
exclusively  to  psychology  and  to  offer  facilities  for  train- 
ing students  in  the  methods  of  the  new  science.  Students 
from  many  parts  of  the  world  were  attracted  to  Wundt's 
laboratory,  and  upon  their  return  founded  laboratories 
in  their  own  countries. 


The  problems  investigated  in  these  early  laboratories 
testify  to  the  close  kinship  of  experimental  psychology 
with  physiology.  The  psychology  of  visual  and  auditory 
sensation,  reaction  time,  psychophysics,  and  association 
constituted  nearly  the  entire  field  of  experimentation. 
It  was  characteristic  of  the  early  experimental  psycholo- 
gists either  to  ignore  individual  differences  or  to  regard 
them  in  the  nature  of  experimental  "errors."  The  greater 
the  individual  variation  in  a  phenomenon,  the  less  ac- 
curate would  be  the  general  laws  discovered  in  regard  to 
it.  Thus  the  extent  of  individual  differences  represented 
the  "  probable  error "  to  be  expected  in  the  operation  of 
the  general  laws  whose  formulation  was  the  only  aim  of 
the  experimental  psychology  of  this  period. 

It  is  apparent  that  the  rise  of  experimental  psychology 
shifted  the  emphasis  away  from  the  study  of  individual 
differences  rather  than  towards  it.  Its  one  contribution 
to  the  development  of  a  differential  psychology  is  to  be 
found  in  its  demonstration  that  psychological  phenomena 
are  amenable  to  objective  and  even  quantitative  investiga- 
tion, that  psychological  theories  can  be  tested  by  actual 
data,  that  psychology,  in  short,  could  become  an  empirical 
science.  Such  a  step  was  required  before  theories  about 
the  individual  could  be  replaced  by  studies  on  individual 


With  the  spread  of  Darwinism  in  the  late  nineteenth 
century,  psychology  became  consistently  more  biological 
in  its  approach.  One  of  the  most  widely  known  of  Dar- 
win's followers  was  Sir  Francis  Galton,  who  first  attempted 
to  apply  the  evolutionary  principles  of  variation,  selection, 
and  adaptation  to  the  study  of  human  individuals.  Gal- 
ton's  scientific  pursuits  were  many  and  varied,  but  they 


were  unified  by  an  underlying  interest  in  the  study  of 
heredity.  The  science  of  Eugenics,  whose  aim  is  the 
control  and  direction  of  human  evolution,  was  originated 
by  Gal  ton.  In  1869,  he  published  a  book  entitled  Hmdi** 
tary  Genius,  in  which,  by  the  application  of  the  now 
well-known  family  history  method,  he  tried  to  demon- 
strate the  inheritance  of  specific  talents  in  various  fields 
of  work.  In  connection  with  the  study  of  human  inherit- 
ance, it  soon  became  apparent  that  related  and  unrelated 
individuals  must  be  measured,  objectively  and  in  large 
numbers,  in  order  to  discover  the  degrees  of  resemblance 
among  them.  For  this  purpose,  Galton  devised  numerous 
tests  and  measures  and  in  1882  established  his  famous 
anthropometric  laboratory  at  South  Kensington  Museum 
in  London.  There,  for  the  payment  of  a  small  fee,  indi- 
viduals could  be  tested  in  sensory  discrimination,,  motor 
capacities,  and  other  simple  processes. 

Through  the  measurement  of  sensory  processes,  Galton 
hoped  to  arrive  at  an  estimate  of  the  subject's  intellectual 
level.  In  the  Inquiries  into  Human  Faculty,  a  collection 
of  miscellaneous  essays  published  in  1883,  he  wrote:  "The 
only  information  that  reaches  us  concerning  outward 
events  appears  to  pass  through  the  avenue  of  our  senses; 
and  the  more  perceptive  the  senses  are  of  difference,  the 
larger  is  the  field  upon  which  our  judgment  and  intelligence 
can  act"  (9,  p.  27).  And  again,  on  the  basis  of  findings 
on  the  inferior  sensitivity  of  idiots,  he  observes  that  this 
sensory  discriminative  capacity  "would  on  the  whole  be 
highest  among  the  intellectually  ablest*'  (9,  p.  29).  For 
this  reason,  measures  of  sensory  capacity,  such  as  vision 
and  hearing,  constituted  a  relatively  large  portion  of  the 
tests  which  Galton  constructed  and  employed.  Among 
these  tests  may  be  mentioned  the  Galton  bar  for  visual 
discrimination  of  length,  the  Galton  whistle  for  the  de- 


termination  of  the  highest  audible  pitch,  kinsesthetic  dis- 
crimination tests  based  on  the  arrangement  of  a  series 
of  weights,  as  well  as  tests  of  strength  of  movement,  speed 
of  simple  reactions,  and  many  others  of  a  similar  nature. 

Galton  also  initiated  the  use  of  "free  association"  tests, 
a  technique  which  was  subsequently  adopted  and  followed, 
up  by  Wundt.  Galton's  study  of  mental  imagery  is  well 
known  and  represents  the  first  extensive  psychological 
use  of  questionnaire  methods.  In  this  questionnaire, 
the  subject  was  directed  "to  think  of  some  definite  ob- 
ject-— suppose  it  is  your  breakfast  table  as  you  sat  down 
to  it  this  morning — and  consider  carefully  the  picture 
that  rises  before  your  mind's  eye"  (9,  p.  84).  They  were 
then  to  describe  the  picture  with  reference  to  illumination, 
definition,  and  coloring.  Wide  individual  and  group 
differences  were  revealed  by  this  analysis  of  imagery. 

A  further  and  very  significant  contribution  of  Galton 
to  differential  psychology  was  his  development  of  statis- 
tical methods  for  the  analysis  of  the  data  of  individual 
differences.  Formerly,  statistics  had  been  chiefly  the 
tool  of  the  trained  mathematician  and  the  professional 
gambler.  Statistical  techniques  were  not  available  in  a 
form  which  would  enable  the  mathematically  untrained 
worker  in  the  biological  sciences  to  employ  them.  Galton 
realized  the  need  for  such  techniques  and  developed  many 
of  the  statistical  procedures  in  current  use  today.  This 
phase  of  his  work  has  been  extended  and  increased  in 
scope  by  many  eminent  students,  chief  among  whom  is 
Karl  Pearson,  who  succeeded  Galton  as  director  of  the 
anthropometric  laboratory  in  1911. 


The  term  "mental  test"  was  first  employed  in  1890 
by  Cattell,  in  an  article  entitled  Mental  Tests  and  Measure- 


menls  (6).  James  McKeen  Cattell  was  an  American 
student  of  Wundt.  In  1888,  having  obtained  his  doctorate 
at  Leipzig,  he  returned  to  this  country,  where  he  was  in- 
strumental both  in  the  spread  of  experimental  psychology 
and  in  the  development  of  mental  testing.  He  was 
influenced  by  Galton's  work  in  test  construction  and 
statistics.  In  Cattell,  we  find  a  convergence  of  two  con- 
temporary movements  in  psychology,  the  rise  of  the  ex- 
perimental method  and  the  measurement  of  individual 
differences.  It  was  characteristic  of  all  of  the  early 
American  mental  tests  that  they  developed  in  the  psy- 
chological laboratory  and  partook  of  the  nature  of  the 
experimental  psychology  of  the  time.  This  was  not  true 
of  many  of  the  tests  developed  in  other  countries. 

In  addition  to  his  experiments  on  reaction  time,  atten- 
tion span,  controlled  association,  reading,  psychophysics, 
and  similar  problems,  Cattell  constructed  a  series  of  tests 
which  were  administered  for  many  years  to  freshmen  and 
seniors  at  Columbia  College.  This  series  included  the 
following  tests:1  (i)  strength  of  grip;  (2)  rate  of  arm 
movement;  (3)  two-point  threshold  on  the  back  of  the 
hand;  (4)  amount  of  pressure  required  to  produce  pain 
on  the  forehead;  (5)  least  noticeable  difference  in  weights; 
(6)  reaction  time  to  sound;  (7)  speed  of  color  naming; 
(8)  bisection  of  a  50-011.  line;  (9)  reproduction  of  a  10 
second  time  interval;  (10)  auditory  memory  span  for 
letters.  This  list  is  reproduced  in  full  since  it  is  typical 
of  the  various  test  series  which  appeared  at  the  time. 

In  1891,  Miinsterberg  (17)  described  a  series  of  tests 
which  he  had  employed  on  schoolchildren.  Tests  of 
reading,  controlled  association  of  various  sorts,  judgment, 
memory,  and  other  simple  mental  processes  were  included. 
At  the  1893  Columbian  Exposition  at  Chicago,  Jastrow 

1  For  a  fuller  description,  cf.  Cattell  and  Farrand  (7). 


administered  a  series  of  sensory,  motor,  and  simple  per- 
ceptual tests  to  all  persons  interested.  Norms  of  physical 
growth  and  mental  development  were  presented  with  the 
tests  (cf.  20). 

What  is  probably  the  first  attempt  to  evaluate  test 
scores  in  terms  of  an  independent  criterion  is  to  be  found 
in  the  study  by  Bolton  (4)  reported  in  1892.  Bolton 
analyzed  data  collected  by  Boas  on  about  1500  school- 
children. The  children's  memory  spans  were  compared 
with  their  teachers'  estimates  of  "  intellectual  acuteness," 
very  little  correspondence  being  found.  Gilbert  (10), 
in  1893,  compared  teachers'  estimates  of  " general  ability" 
on  some  1200  children  with  their  scores  on  eight  tests 
of  sensory  and  motor  functions,  reaction  time,  sensory 
memory,  and  suggestibility.  Three  years  later,  Gilbert  (n) 
described  some  additional  tests  and  reported  the  results 
obtained  with  them  on  several  hundred  children.  The 
data  were  analyzed  in  respect  to  sex  differences,  intel-, 
lectual  growth,  and  the  relationship  of  mental  and  physical 

In  Germany,  Oehrn  (19),  a  pupil  of  Kraepelin,  pub- 
lished in  1889  the  results  of  an  intensive  study  of  a  series 
of  tests  on  ten  subjects.  The  tests  had  been  rather  arbi- 
trarily selected  to  measure  perception,  memory,  associa- 
tion, and  motor  functions.  In  1895,  Kraepelin  (15)  pro- 
posed a  set  of  traits  which  he  regarded  as  basic  in  the 
characterization  of  any  individual.  He  also  devised  tests 
for  the  measurement  of  these  traits,  most  of  the  tests 
involving  simple  arithmetic  operations.  These  tests  were 
of  rather  dubious  validity  for  measuring  the  traits  in 
question,  and  in  addition  they  were  quite  impracticable, 
some  of  them  requiring  several  days  for  their  completion. 
Research  on  mental  tests  was  also  being  conducted  simul- 
taneously under  the  direction  of  the  Italian  psychologist 


Ferrari.  In  an  article  appearing  in  1896,  some  of  these 
tests  were  described  (12).  They  included  measures  of 
vasomotor  activity,  motor  strength  and  skill,  range  of 
apprehension,  description  of  pictures,  and  temporal  esti- 
mation. Interesting  individual  differences  were  reported 
in  many  of  these  tests. 


At  the  turn  of  the  century,  differential  psychology  had 
begun  to  assume  very  definite  shape.  In  1895,  Binet  and 
Henri  published  an  article  entitled  La  psychologie  indi- 
vidudle  (3),  which  represents  the  first  systematic  analysis 
of  the  aims,  scope,  and  methods  of  differential  psychology. 
Their  opening  sentence  suggests  the  status  of  this  branch 
of  psychology  at  the  time.  It  reads:  "We  broach  here 
a  new  subject,  difficult  and  as  yet  very  meagerly  explored" 
(p.  411).  Binet  and  Henri  put  forth  as  the  two  major 
problems  of  differential  psychology,  first,  the  study  of  the 
nature  and  extent  of  individual  differences  in  psychologi- 
cal processes;  and  secondly,  the  discovery  of  the  interrela- 
tionships of  mental  processes  within  the  individual,  so  that 
we  may  arrive  at  a  classification  of  traits  and  determine 
which  are  the  more  basic  functions. 

In  1900  appeared  the  first  edition  of  Stern's  book  on 
differential  psychology,  under  the  title  Uber  Psychologie 
der  indimduellen  Differ  enzen  (25).  Part  I  deals  with  the 
nature,  problems,  and  methods  of  differential  psychology. 
Within  the  scope  of  this  branch  of  psychology  Stern 
included  differences  among  individuals  as  well  as  among 
racial  and  cultural  groups,  occupational  and  social  levels, 
and  the  two  sexes.  The  fundamental  problem  of  differ- 
ential psychology  he  characterized  as  threefold:  (i)  What 
is  the  nature  and  extent  of  differences  in  the  psychological 
life  of  individuals  and  groups?  (2)  What  factors  deter- 


mine  or  affect  these  differences? — in  this  connection  he 
mentioned  heredity,  climate,  social  or  cultural  level,  train- 
ing, adaptation,  etc.  (3)  How  are  the  differences  mani- 
fested? Can  they  be  detected  by  such  indices  as  hand- 
writing, facial  conformation,  etc.?  Stern  also  included  a 
discussion  of  the  concepts  of  psychological  type,  individu- 
ality, and  normality  and  abnormality.  Under  methods 
of  differential  psychology,  he  gave  an  evaluation  of  intro- 
spection, objective  observation,  the  use  of  material  from 
history  and  poetry,  the  study  of  culture,  quantitative 
testing,  and  experiment.  Part  II  contains  a  general 
discussion  and  certain  data  on  individual  differences  in 
various  psychological  traits,  from  simple  sensory  capac- 
ities to  more  complex  mental  processes  and  emotional 
characteristics.  Stern's  book  appeared  in  a  highly  re- 
vised and  enlarged  edition  in  1911,  and  again  in  1921, 
under  the  title  of  Differentielle  Psychologic  in  ihren  metho~ 
dischen  Grundlagen  (26). 

In  America,  committees  were  being  appointed  to  in- 
vestigate testing  methods  and  to  sponsor  the  accumula- 
tion of  data  on  individual  differences.  At  its  1895  meet- 
ing, the  American  Psychological  Association  appointed  a 
committee  "to  consider  the  feasibility  of  cooperation 
among  the  various  psychological  laboratories  in  the  col- 
lection of  mental  and  physical  statistics "  (7,  p.  619).  In 
the  following  year,  the  American  Association  for  the 
Advancement  of  Science  established  a  standing  committee 
to  organize  an  ethnographic  survey  of  the  white  population 
in  the  United  States.  Cattell,  one  of  the  members  of  this 
committee^  pointed  out  the  importance  of  including  psy- 
chological tests  in  this  survey  and  suggested  that  its 
work  be  coordinated  with  that  proposed  by  the  American 
Psychological  Association  (7,  pp.  619-620). 

The  application  of  the  newly  devised  mental  tests  to 


various  groups  was  also  getting  under  way,  R.  L.  Kelly 
(13)  In  1903  and  Norsworthy  (18)  In  1906  compared 
normal  and  feebleminded  children  on  sensori-motor  and 
simple  mental  tests,  and  called  attention  to  the  continuous 
gradation  in  ability  which  exists  between  these  groups, 
the  feebleminded  not  constituting  a  distinct  category.  In 
1903  appeared  Thompson's  The  Mental  Traits  of  Sex  (28), 
the  result  of  several  years'  testing  of  men  and  women 
with  a  variety  of  tests.  This  represents  the  first  compre- 
hensive investigation  on  sex  differences.  Tests  of  sensory 
acuity,  motor  capacities,  and  a  few  simple  mental  proc- 
esses were  also  being  administered  for  the  first  time  to 
various  racial  groups.  A  few  scattered  investigations  ap- 
peared before  1900.  In  1904  Woodworth  (30)  and  Bruner 
(5)  tested  several  primitive  groups  at  the  St.  Louis  Ex- 
position. In  the  same  year  appeared  Spearman's  original 
article  putting  forth  his  Two-Factor  theory  of  mental  or- 
ganization and  introducing  a  statistical  technique  for  In- 
vestigating the  problem.  Thus,  shortly  after  the  begin- 
ning of  the  century,  the  foundations  had  been  laid  for 
every  branch  of  differential  psychology. 


The  intelligence  test  Is  a  product  of  the  twentieth 
century.  The  early  mental  tests  were  predominantly 
sensori-motor  or  very  simple  in  nature.  This  was  no 
doubt  a  carry-over  from  the  sensationism  current  in  the 
psychological  laboratories  of  the  time.  Complex  mental 
processes  were  believed  to  be  best  understood  by  an- 
alyzing them  into  their  elementary  components,  usually 
of  a  sensory  nature.  Most  of  the  efforts  of  the  early 
experimentalists  were  therefore  devoted  to  the  study  of 
sensations  and  simple  reactions,  and  this  influence  left 
its  mark  on  the  newly  developing  mental  tests. 


Binet  and  Henri,  in  their  1895  article  (3),  were  the  first 
definitely  to  point  out  the  need  for  more  complex  tests 
to  measure  " intelligence."  They  examined  the  five  most 
comprehensive  current  test  series,  those  of  Cattell,  Miin- 
sterberg,  Jastrow,  Kraepelin,  and  Gilbert,  and  found  all 
of  them  greatly  overweighted  with  sensory  tests  and 
lacking  in  tests  of  complex  processes.  From  an  analysis 
of  the  available  data,  they  concluded  that  individual 
differences  are  more  marked  in  complex  tasks  and  that 
the  latter  are  therefore  better  suited  to  the  study  of  such 
differences.  Partly  to  remedy  this  deficiency  in  the  cur- 
rent tests,  Binet  and  Henri  described  ten  types  of  tests 
which  in  their  opinion  would  yield  the  largest  and  most 
significant  individual  differences.  The  series  included 
tests  of  memory,  mental  imagery,  imagination,  attention, 
comprehension,  suggestibility,  aesthetic  appreciation,  moral 
feelings,  muscular  force  and  force  of  will,  and  motor  abil- 
ity and  visual  discrimination.  The  entire  series,  according 
to  the  authors,  would  require  only  from  one  to  one  and 
one-half  hours. 

In  1897,  Ebbinghaus  (8)  proposed  a  theory  to  the  effect 
that  intelligence  is  the  ability  to  combine  or  integrate 
the  items  of  experience,  and  offered  the  sentence  com- 
pletion test  as  a  technique  for  measuring  this  ability. 
In  this  test,  the  subject  is  presented  with  sentences  in 
which  certain  of  the  words  are  missing  and  he  is  required 
to  fill  in  the  proper  words.  In  experiments  on  German 
schoolchildren,  Ebbinghaus  had  found  this  test  more  ef- 
fective than  simpler  tests  of  calculation  and  memory. 
The  completion  test  showed  the  most  regular  increase  in 
score  with  age  and  it  was  also  the  only  one  of  the  tests 
employed  which  differentiated  clearly  among  those  pupils 
within  each  grade  whose  scholastic  standing  was  good, 
average,  or  poor.  Binet's  contention  for  the  superiority 


of  the  more  complex  tests  In  differential  psychology  was 
+kus  corroborated. 

Two  American  studies  of  this  period  lent  further  sup- 
port to  Binet's  statements.  One  of  these  studies  (22) 
was  conducted  by  Sharp,  a  student  of  Titchener,  and  was 
designed  as  a  specific  investigation  of  the  conclusions  of 
Binet  and  Henri.  A  set  of  tests,  modeled  largely  on  those 
of  Binet  and  Henri,  was  administered  to  seven  advanced 
psychology  students.  The  experiment  was  very  intensive 
and  included  the  repetition  of  similar  tests  on  different 
days  to  determine  the  consistency  of  the  processes  tested. 
In  general,  although  the  need  for  further  controls  and 
refinements  was  suggested,  the  tests  proved  satisfactory 
and  yielded  sizable  individual  differences  in  spite  of  the 
homogeneous  and  select  nature  of  the  group.  Sharp 
concluded:  "We  concur  with  Mm.  Binet  and  Henri  in 
believing  that  individual  psychical  differences  should  be 
sought  for  in  the  complex  rather  than  in  the  elementary 
processes  of  mind,  and  that  the  test  method  is  the  most 
workable  one  that  has  yet  been  proposed  for  investigat- 
ing these  processes"  (22,  p.  390).  A  few  years  later, 
Wissler  (29)  published  the  results  of  his  correlation  anal- 
ysis of  the  data  collected  in  CattelPs  laboratory.  The 
correlations  showed  "little  more  than  a  chance  relation " 
among  the  tests,  and  also  a  negligible  correspondence 
with  academic  grades.  Thus  the  inadequacy  of  the  simple 
tests  originally  employed  was  again  indicated. 

Against  this  background  of  theory  and  data  appeared 
the  first  intelligence  scale.  In  1904  the  French  Minister 
of  Public  Instruction  appointed  a  committee  to  investi- 
gate the  causes  of  retardation  among  public  schoolchil- 
dren. Binet  was  one  of  the  members  of  this  committee. 
As  a  direct  outgrowth  of  his  work  in  this  connection, 
Binet  published,  in  collaboration  with  Simon,  the  1905 


scale  for  measuring  Intelligence.  This  scale  consisted  of 
30  problems  arranged  in  a  rough  order  of  difficulty.  In 
1908  appeared  Binet's  first  revision  of  the  scale,  in  which 
the  tests  were  grouped  into  age  levels  and  the  concept 
of  "mental  age"  1  was  introduced.  The  scale  was  again 
revised  in  1911,  the  year  of  Binet's  untimely  death.2 

The  Binet  tests  have  been  translated  into  more  than  a 
dozen  languages  and  their  use  has  spread  over  every  con- 
tinent. In  America,  five  different  revisions  have  appeared, 
of  which  the  most  widely  known  are  the  Stanford-Binet, 
prepared  by  Terman  and  his  associates  at  Stanford 
University,  and  the  Kuhlman-Binet  which  extended  the 
scale  down  to  the  three-months  age  level.  The  intelligence 
quotient  (I.Q.)?  found  by  dividing  the  child's  mental 
age  by  his  chronological  age,  was  first  employed  in  the 
Stanford-Binet,  although  its  use  and  advantages  had  been 
previously  discussed  by  Stern  and  others.  The  develop- 
ment of  performance  scales  for  testing  the  deaf,  the 
illiterate,  and  the  foreign,  to  whom  language  tests  are 
inapplicable,  may  be  mentioned  as  a  further  development 
of  intelligence  testing. 


The  Binet  scale  and  its  revisions  are  "individual  tests" 
in  the  sense  that  only  one  subject  can  be  tested  at  a  time. 
Furthermore,  owing  to  the  nature  of  these  tests,  a  highly 
trained  examiner  is  required  to  administer  them.  Testing 

1The  child's  score  on  an  age  scale  is  expressed  as  a  mental  age  (M.A.).  If, 
for  example,  he  passes  successfully  all  of  the  tests  assigned  to  the  loyear  level, 
he  has  a  mental  age  of  10,  regardless  of  what  his  chronological  age  may  be. 

2  Attention  has  recently  been  called  to  the  fact  that  as  early  as  1887  a  series 
of  developmental  standards  and  simple  tests  for  judging  the  mental  level  of 
infants  during  the  first  three  years  was  worked  out  by  an  American  physician, 
Dr.  S.  E.  Chaille.  The  concept  of  mental  age  seems  to  have  been  implicit  in  his 
treatment  of  the  data,  although  this  term  was  not  employed  (cf.  Goodenough, 
F.  L.,  "An  Early  Intelligence  Test,"  Child  Dev.,  1934, 5, 13-18). 


on  a  large  scale  could  not  be  carried  on  under  these 
conditions.  Data  on  such  problems  as  sex  and  race  differ- 
ences, for  example,  which  require  the  investigation  of  large 
samples,  would  be  very  slow  in  accumulating. 

The  advent  of  the  group  intelligence  scale  was  probably 
the  chief  factor  in  bringing  about  the  popular  mental 
testing  fad.  The  group  test  is  designed  with  a  view  to  its 
general  use.  It  is  not  only  adapted  to  the  simultaneous 
testing  of  large  groups,  but  it  is  also  relatively  easy  to 
administer  and  fool-proof  in  its  scoring.  The- impetus 
for  the  development  of  group  tests  was  given  by  the 
pressing  need  of  testing  over  one  and  one-half  million 
men  in  the  United  States  army  in  1917.  A  quick,  rough 
classification  in  respect  to  intelligence  was  necessary  for 
many  purposes.  Discharge  because  of  serious  mental 
defect,  assignment  to  labor  battalions  requiring  only  low- 
grade  work,  admittance  to  officers'  training  camps,  and 
a  number  of  similar  problems  required  a  knowledge  of 
the  intellectual  level  of  the  soldier. 

Accordingly,  a  committee  was  appointed  by  the  Amer- 
ican Psychological  Association  to  devise  a  test  suited  to 
this  purpose.  The  committee  consisted  of  five  psycholo- 
gists who  were  specialists  in  mental  testing,  and  was 
under  the  direction  of  Robert  M.  Yerkes.  All  of  the 
available  material  on  mental  tests  was  examined  for  its 
suitability  to  the  needs  of  the  army  testing  program.  The 
greatest  assistance  was  derived  from  an  unpublished  group 
scale  which  had  previously  been  developed  by  Otis,  and 
which  the  latter  made  available  to  the  government.  The 
final  outcome  of  the  research  of  the  army  psychologists 
was  the  Army  Alpha  and  the  Army  Beta.  The  former 
was  the  more  widely  used  of  the  two;  the  latter  is  a  non- 
language  scale,  and  was  designed  for  testing  illiterates 
and  foreigners  unfamiliar  with  English. 


After  the  close  of  the  World  War,  group  tests  were 
constructed  at  a  rapid  rate.  Soon  special  tests  were 
available  for  elementary  schoolchildren  as  well  as  kinder- 
garten and  pre-school  levels,  high  school  and  college  stu- 
dents, and  unselected  adults.  Mental  testing  attained 
heretofore  undreamed-of  proportions.  School  teachers 
were  now  qualified  to  administer  the  newly  simplified 
tests;  large-scale  school  surveys  were  initiated;  college 
freshmen  were  tested  as  part  of  the  routine  of  admission; 
the  general  public  became  intelligence-test-conscious,  and 
the  I.Q.  became  a  byword  in  everyday  conversation. 

This  sudden  popularization  and  publicity  was  an  un- 
fortunate handicap  in  the  development  of  a  measuring 
instrument  still  in  its  infancy.  The  intelligence  scales 
were  as  yet  very  crude  when  they  were  put  into  the  hands 
of  the  layman.  They  were  too  often  accepted  as  a  finished 
product  and  an  infallible  guide.  Analysis  of  results  and 
evaluation  of  techniques  were  subordinated  to  the  more 
alluring  occupation  of  classifying  people.  Occasionally 
psychologists  themselves  were  guilty  of  over-hasty  gen- 
eralization. Data  on  the  various  problems  of  differential 
psychology  were  being  amassed  in  a  rush.  Sweeping 
conclusions  were  drawn — and  quoted. 


The  extension  of  testing  techniques  from  sensory  and 
intellectual  capacities  to  emotional  and  social  traits  is 
also  a  very  recent  development.  An  antecedent  of  current 
personality  testing  may  be  found  in  Kraepelin's  first  use 
of  the  free  association  test  on  pathological  cases  and  on 
individuals  who  had  been  experimentally  subjected  to 
various  influences  such  as  fatigue,  hunger,  and  drugs. 
Kraepelin  (14)  reported  that  all  of  these  agencies  increased 
the  number  of  superficial  associations.  In  1894,  Sommer 


(23)  suggested  that  mental  disorders  could  be  differenti- 
ated by  means  of  the  free  association  test.  The  use  of  this 
test,  for  a  variety  of  purposes,  has  persisted  to  the  present. 

The  most  fruitful  approach  to  the  measurement  of 
personality  traits,  however,  has  proved  to  be  through 
standardized  questionnaire  and  rating  scale  methods. 
These  methods  were  originally  developed  by  Gallon, 
Pearson,  and  Cattell  in  a  different  connection.  The 
first  widely  employed  personality  questionnaire  was  the 
Woodworth  "Personal  Data  Sheet"  (cf.  27,  Ch.  V),  an 
inventory  of  neurotic  tendencies  and  emotional  malad- 
justments devised  for  use  on  the  American  soldiers  during 
the  World  War.  This  questionnaire  was  prepared  by 
Woodworth,  who  was  chairman  of  the  Committee  on 
Emotional  Fitness  appointed  by  the  National  Research 
Council.  Although  the  armistice  came  before  the  final 
form  of  the  questionnaire  could  be  extensively  applied, 
it  was  used  subsequently  in  army  hospitals  and  yielded 
much  interesting  information. 

Several  revisions  and  adaptations  of  the  Woodworth 
questionnaire  have  appeared,  including  forms  especially 
suited  for  children,  and  for  college  students.  Tests  have 
also  appeared  for  the  measurement  of  various  other  so- 
cial and  emotional  characteristics,  such  as  introversion- 
extroversion,  ascendance-submission,  and  self-sufficiency. 
A  fairly  recent  development  is  the  construction  of  objec- 
tive " performance"  tests  of  moral  or  character  traits, 
such  as  honesty,  cooperation,  and  self-control.  Data  on 
individual  and,  to  a  lesser  extent,  group  differences  are 
being  steadily  accumulated  with  all  of  these  tests. 


After  about  twenty  years  of  phenomenal  development 
in  the  construction  and  application  of  mental  tests,  psy- 


chologists  are  beginning  to  turn  their  attention  to  a 
critical  analysis  of  testing  techniques  themselves.  There 
is  apparent  a  growing  tendency  to  question  generaliza- 
tions, to  scrutinize  the  conditions  under  which  data  were 
obtained,  and  to  evaluate  the  results  in  the  light  of  such 
conditions.  A  reaction  is  setting  in  against  the  indiscrimi- 
nate use  of  mental  tests  which  followed  the  rise  of  group 
testing.  In  differential  psychology,  this  reaction  is  mani- 
fested in  two  chief  ways.  In  the  first  place,  increasing 
stress  is  being  laid  upon  the  influence  of  environmental 
conditions  in  determining  or  modifying  the  individual's 
characteristics.  In  the  second  place,  the  measurement  of 
"general  intelligence"  is  losing  its  prominence  as  the  goal 
of  testing.  Individual  as  well  as  group  differences  are 
being  sought  in  narrower  and  more  clearly  definable  traits. 

The  original  aim  of  the  mental  testers  was  the  measure- 
ment of  the  individual's  "capacities"  or  potentialities 
of  intellectual  development,  as  distinguished  from  his 
present  skills  and  information.  The  measurement  of  the 
latter  would  have  been  a  relatively  simple  task.  If  we 
wish  to  ascertain  whether  an  individual  is  proficient  in 
many  languages,  for  example,  we  need  only  examine  his 
knowledge  of  all  languages  with  which  he  claims  familiar- 
ity. But  if  we  want  to  know  whether  this  individual  can 
learn  languages  easily,  whether  it  would  be  worth  while 
to  teach  him,  or  to  advise  him  to  enter  a  vocation  which 
demands  a  mastery  of  several  languages,  then  we  are 
faced  with  a  much  more  difficult  problem.  This  was  the 
type  of  problem  with  which  the  mental  tester  tried  to  cope. 

If  one  is  to  determine  what  the  individual  can  do  rather 
than  what  he  has  already  accomplished,  it  was  argued, 
it  is  necessary  in  some  way  to  "rule  out"  differences  in 
the  previous  experience  or  training  of  the  subjects.  This 
was  attempted  either  by  presenting  material  which  was 


equally  unfamiliar  to  all,  or  by  the  reverse  procedure  of 
utilizing  only  material  common  to  everyone's  experience. 
Frequently,  the  two  methods  were  combined,  as  in  the 
use  of  familiar  material  in  a  novel  and  unusual  manner. 
Such  a  procedure  is  a  practicable  one  and  will  yield  useful 
information,  provided  its  limitations  and  assumptions  are 
kept  clearly  in  view.  In  respect  to  the  use  of  familiar 
material,  it  becomes  necessary  to  ascertain  that  the  ma- 
terial is  actually  familiar,  in  an  approximately  equal 
degree,  to  all  of  the  subjects.  A  test  is  applicable  only 
to  individuals  similar  in  their  experiential  background 
to  the  group  upon  whom  it  was  standardized.  When 
given  to  individuals  from  different  national  or  cultural  groups ', 
or  from  widely  differing  economic,  social,  or  educational 
levels,  "intelligence  tests"  do  little  more  than  reflect  the 
varied  backgrounds  of  the  subjects.  Under  such  conditions, 
the  tests  fail  to  fulfill  the  original  purpose  for  which  they 
were  constructed.  We  are  here  concerned  only  in  present- 
ing this  point  of  view  as  a  recent  historical  development; 
further  elaboration  of  its  theoretical  implications  will  be 
found  in  subsequent  chapters. 

The  current  trend  toward  the  development  of  tests  of 
separate  traits  and  special  aptitudes  likewise  results  from 
a  more  searching  analysis  of  testing  methods.  Recent 
research  on  the  relationships  among  abilities  suggests 
that  important  individual  differences  might  be  obscured 
if  all  abilities  are  lumped  indiscriminately  under  the  in- 
clusive concept  of  "general  intelligence."  From  the  prac- 
tical standpoint  of  prediction,  tests  of  special  aptitudes 
are  being  constructed  to  fit  the  needs  of  particular  educa- 
tional or  vocational  situations.  From  the  standpoint  of 
theoretical  interpretation,  investigations  on  independent 
traits  or  unitary  abilities  are  contributing  towards  a  clearer 
understanding  of  individual  and  group  differences. 



1.  Aristotle.    The  Works  of  Aristotle^  ed.  by  W.  D.  Ross.    Ox- 
ford: Clarendon  Press,   1915.     Vol.  IX,  Magna  Moralia, 
Ethica  Eudemia,  de  Virtutibus  et  Vittis. 

2.  Bain,  Alexander.     The  Senses  and  the  Intellect.     London: 
Parker,  1855.  Pp-  614. 

3.  Binet,  A.,  and  Henri,  V.     "La  psychologic  individuelle," 
Annee  fsychoL,  1895,  2,  411-463. 

4.  Bolton,  T.  L.   "The  Growth  of  Memory  in  Schoolchildren," 
Amer.  J.  Psychol.,  1891-92,  4,  362-380. 

5.  Bruner,  F.  G.    "The  Hearing  of  Primitive  Peoples,"  Arch. 
Psychol.,  1908,  No.  n.  Pp.  113. 

6.  Cattell,  J.  McK.  "Mental  Tests  and  Measurements,"  Mind, 
1890,15,  373"38&. 

7.  Cattell,  ].  McK.,  and  Farrand,  L.    "Physical  and  Mental 
Measurements  of  the  .Students  of  Columbia  University," 
Psychol.  Rev,,  1896,  3,  618-648. 

8.  Ebblnghaus,  H.   "tJber  eine  neue  Methode  zur  Priifung 
geistiger   Fahigkeiten  und  ihre  Anwendung  bei  Schulkin- 
dern/'  Zscht.  f.  angew.  Psychol.,  1897,  13,  401-459. 

9.  Galton,  F.    Inquiries  into  Human  Faculty  and  Its  Develop- 
ment. London:  Macmillan,  1883.   Pp.387. 

10.  Gilbert,  J.  A.     "Researches  on  the  Mental  and  Physical 
Development  of  School  Children,"  Stud.  Yale  Psychol.  Lab., 
1894,  2,  40-100. 

11.  -  .     "Researches  upon  Children  and  College  Students," 
Iowa  Univ.  Stud.  Psychol^  1897,  I,  1-39. 

12.  Guiccardi,  G.,  and  Ferrari,  G,  C.  "I  testi  mentali  perl'esame 
degli  alienati,"  Riv.  sper.  di  freniatria,  1896,  22,  297-314. 

13.  Kelly,  R.  L.    "Psychophysical  Tests  of  Mentally  Deficient 
Children,"  Psychol.  Rev*,  1903,  10,  34S~372- 

14.  Kraepelin,  E.    Uber  die  BeeinfiHssung  einjacher  pychischer 
Forgange  durch  einige  ArmeimitteL     Jena:  Fischer,  1892, 
Pp.  259. 

j^  -  m     "Der  psychologische  Versuch  in  der  Psychiatric," 
Psychol.  Arbeiten,  1895,  I,  1-91. 


16.  Monroe,  P.   A  Textbook  in  the  History  of  Education.   N.  Y.: 
Macmillan,  1926.  Pp.  759. 

17.  Miinsterberg,  EL   "Zur  Individualpsychologie,"  Centralblatt 
fur  Nervenheilkunde  und  Psychiatrie,  1891,  14,  196-198. 

1 8.  Nors worthy,  N.    "The  Psychology  of  Mentally  Deficient 
Children/'  Arch.  PsychoL,  1906,  No.  i.  Pp.  in. 

19.  Oehrn,  A.  Experimented  Studien  zur  Individualpsychologie. 
Dorpater  Dissertation,  1889  (also  publ.  in  PsychoL  Arbeiten^ 

20.  Philippe,    J.       "Jastrow — Exposition    d'anthropologie    de 
Chicago — Tests  psychologiques,  etc.,"  Annee  PsychoL,  1894, 

21.  Plato.    The  Republic  of  Plato  (transl.  by  J.  L.  Davies  and 
D.  J.  Vaughan).  N.  Y.:  Burt,  19—.  Pp.  406. 

22.  Sharp,  S.  E.  "Individual  Psychology:  a  Study  in  Psycholog- 
ical Method,"  Amer.  J.  PsychoL,  1898-99,  10,  329-391. 

23.  Spmmer,  R.  Diagnostik  der  Geisteskrankheiten  fur  prakiische 
Arzte  und  Siudierende.     Wien  und  Leipzig:  Urban  und 
Schwarzenberg,  1894.  Pp.  302. 

24.  Spearman,  C.  " * General  Intelligence'  Objectively  Deter- 
mined and  Measured,"  Amer.  J.  PsychoL^  1904,  15,  201-293. 

25.  Stern,  W.     Uber  Psychologie  der  indwiduellen  Diffmnzen 
(Ideen  zur  einer  " Differ enzdlen  Psychologic"}*      Leipzig: 
Earth,  1900.  Pp.  146. 

26.  .     Die  Differentielle  Psychologie  in  ihren  methodischen 

Grundlagen,  Leipzig:  Earth,  1921.  Pp.  545. 

27.  Symonds,   P.  M.      Diagnosing  Personality   and  Conduct. 
N.  Y.:  Century,  1931.  Pp.  602. 

28.  Thompson,  H.  B.  The  Mental  Traits  of  Sex.  Chicago:  Univ. 
Chicago  Press,  1903.  Pp.  188. 

29.  Wissler,  C.     "The  Correlation  of  Mental  and  Physical 
Tests,"  PsychoL  Mon.,  1901,  3,  No.  16.  Pp.  62. 

30.  Woodworth,  R.  S.    "Race  Differences  in  Mental  Traits," 
Science,  N.  S.,  1910,  31,  171-186. 



Individual  differences  are  quantitative  rather  than  qual- 
itative. Popular  opinion  frequently  classifies  people  in 
reference  to  the  possession  or  non-possession  of  certain 
traits.  Thus  one  individual  is  said  to  have  a  talent  for 
music,  another  for  painting,  a  third  for  mathematics,  a 
fourth  for  organizing  people.  Such  a  characterization, 
however,  results  from  purely  practical  considerations.  In 
order  to  choose  music  as  a  vocation,  or  even  as  a  serious 
avocation,  for  example,  an  individual  must  have  a  certain 
minimum  of  musical  talent;  if  his  degree  of  musical  abil- 
ity falls  below  that  minimum,  he  is  not  regarded  as  "a 
musical  person."  Furthermore,  if  such  an  individual  has 
more  outstanding  ability  along  some  other  line,  he  will 
be  classified  in  respect  to  the  latter  trait  and  not  on  the 
basis  of  his  musical  aptitude.  Qualitative  distinctions 
of  this  sort  are  made  in  practice  and  are  based  on  arbitrary 
or  socially  determined  criteria  and  limits.  Actually,  how- 
ever, every  individual  can  be  described  in  terms  of  the 
same  traits. 

There  are  no  "alternative"  psychological  traits,  in  the 
sense  that  one  trait  serves  as  a  substitute  for  the  other* 
No  traits  are  mutually  exclusive.  It  might  be  argued, 
however,  that  there  are  certain  characteristics  which  a 
person  may  either  have  or  not  have,  and  that  in  this 
respect  we  may  speak  of  qualitative  differences.  The 
classical  examples  are  loss  of  sensation,  such  as  blindness 
or  deafness.  Here,  it  would  seem,  are  traits  characterized 


by  presence  or  absence:  a  person  can  see  or  he  cannot  see, 
he  can  hear  or  not  hear.  This,  too,  turns  out  to  be  a  purely 
conventional  and  practical  distinction.  Anyone  who  has 
visited  an  institution  for  the  blind  knows  that  there  are 
many  degrees  of  blindness,  and  that  not  all  those  classified 
as  blind  are  totally  blind.  The  everyday  working  definition 
of  blindness  is  any  degree  of  visual  deficiency  too  serious 
to  permit  normal  activity.  The  same  is  obviously  true 
of  deafness  and  any  other  sensory  disorder.  Between 
the  empirically  established  " normal"  vision  or  hearing 
and  what  is  classed  as  blindness  or  deafness  there  is  to 
be  found  a  continuous  gradation  of  minor  deficiencies. 
It  should  be  added  that  the  existence  of  a  trait  in  zero 
degree,  as  in  total  blindness,  is  not  inconsistent  with  the 
quantitative  view  of  individual  differences.  The  latter 
implies  only  that  there  be  intermediate  degrees  rather 
than  simple  presence  or  absence. 


Since  individual  differences  have  been  found  to  be 
quantitative,  we  may  now  ask  how  the  varying  degrees 
of  each  trait  are  distributed  among  people.  Are  individu- 
als scattered  uniformly  over  the  entire  range  or  do  they 
cluster  at  one  or  more  points?  What  are  the  relative  fre- 
quencies with  which  different  degrees  of  a  trait  occur? 
These  questions  can  best  be  answered  by  an  examination 
of  frequency  distributions  and  frequency  graphs. 

Like  all  statistical  devices,  the  frequency  distribution 
is  a  means  of  summarizing  and  organizing  quantitative 
data  in  order  to  facilitate  its  treatment  and  reveal  sig- 
nificant trends.  Scores  on  a  test,  or  any  other  set  of 
measures,  are  grouped  into  class-intervals,  and  the  number 
of  cases  falling  within  each  interval  is  tabulated.  An  ex- 
ample of  a  frequency  distribution  is  given  in  Table  L 


This  shows  the  scores  of  1000  college  students  on  a  simple 
learning  test.  The  scores  range  from  8  to  52  and  have 
been  grouped  into  class-intervals  of  four  points.  The 
advantages  of  such  a  table  over  a  list  of  1000  individual 
scores  are  obvious. 

The  facts  brought  out  by  a  frequency  distribution  can 
be  made  more  vivid  if  presented  pictorially  by  means  of  a 
frequency  graph.  In  Figure  I  are  shown  the  data  of 
Table  I  in  graphic  form.  The  base  line  or  horizontal 
axis  represents  the  scores;  the  vertical  axis  shows  the 
frequency  or  number  of  cases  falling  within  each  class- 




(After  Anastasi,  2,  p.  34) 

























8»i  I 


N  —  1000 

interval.  The  graph  has  been  plotted  in  two  ways,  both 
being  about  equally  common.  One  graph  is  a  frequency 
polygon,  in  which  the  number  of  individuals  within  each 
interval  is  indicated  by  a  point,  centrally  located  in 
respect  to  the  class-interval;  the  successive  points  are 
then  joined  by  straight  lines.  The  other  is  obtained  by 



F  /    \  '    Frequency  Polygon 

280  f-  f\       \\ Histogram 


8?  22° 
Q  200 

-  180 

-  160 
|  140 
I  120 


8-   12-  16-  20-   24-  28-  32-  36-  40-  44-  48-  52- 
11   15   19   23   27   31   35   39   43   47   51   55 


(Data  from  Table  I.) 

erecting  a  column  or  rectangle  over  each  class-interval, 
the  height  of  the  column  depending  upon  the  number  of 
cases  in  that  interval.  This  type  of  graph  is  known  as  a 


The  reader  will  already  have  noticed  certain  character- 
istics of  the  distribution  presented  in  Table  I  and  Fig- 
ure i.  The  majority  of  cases  cluster  in  the  center  of  the 
range  and  as  the  extremes  are  approached  there  is  a 
gradual  and  continuous  tapering  off.  The  curve  shows 
no  gaps  or  breaks;  no  clearly  separated  classes  can  be 
discerned.  The  curve  is  also  bilaterally  symmetrical, 
that  is,  if  it  should  be  divided  by  a  vertical  line  through 
the  center,  the  two  halves  so  obtained  would  be  nearly 
identical.  This  distribution  curve  resembles  the  bell- 


shaped  "  normal  curve/'  the  type  most  commonly  found 
in  the  measurement  of  individual  differences.  The  the- 
oretically determined3  ideal  normal  curve  will  be  found 
in  Figure  2. 

The  concept  of  the 
normal  curve  is  an  old 
one  in  statistics.  It  first 
became  prominent  as  the 
normal  probability  curve. 
The  probability  of  the  Fia  a>  THEORETIC^NORMAL  CURVE. 

occurrence  of  an  event  is 

the  expected  relative  frequency  of  occurrence  of  the  given 
event  in  a  very  large,  or  infinite,  number  of  observations. 
This  probability  is  represented  by  a  ratio  or  fraction, 
the  numerator  of  which  is  the  expected  outcome,  and  the 
denominator  the  total  possible  outcomes.  Thus  the  prob- 
ability or  chances  that  when  two  coins  are  tossed  only 
heads  will  come  up  is  f ,  or  one  out  of  four  possible  occur- 
rences; the  probability  of  one  head  and  one  tail  is  J; 
and  that  of  two  tails,  f .  If  the  number  of  coins  is  increased, 
say  to  100,  so  that  the  number  of  possible  occurrences  or 
combinations  becomes  very  large,  we  can  still  determine 
mathematically  the  chances  of  any  one  combination,  such 
as  all  heads  or  twenty  heads  and  eighty  tails,  occurring. 
These  probabilities,  or  expected  frequencies  of  occurrence, 
can  be  plotted  graphically  by  the  same  method  outlined 
above  for  plotting  scores.  The  curve  obtained  when  the 
number  of  coins  is  very  large  will  be  the  bell-shaped 
normal  probability  curve.  In  Figure  3  are  shown  the 
theoretical  and  obtained  frequencies  for  12  dice  thrown 
4096  times.  In  each  throw,  the  number  of  dice  showing 
a  4,  5,  or  6  spot  uppermost  was  determined.  This  number 
could,  of  course,  vary  from  zero  to  12,  the  total  number 
of  dice  thrown.  The  graph  shows  the  relative  frequency 


1000  r  Theoretical 

_  Observed 



|  600 


I*  400 


8"  200 

01         234        56        789        10      11       12 

Number  of  Dice  Showing  a  4,  5,  or  6  Spot  in  Each  Throw 
IN  4096  THROWS  OF  12  DICE.    (After  Yule,  21,  p.  258.) 

of  each  combination  in  the  total  4096  throws.  It  will  be 
noted  that  there  is  a  very  close  agreement  between  the 
theoretical  and  obtained  curves. 

The  results  obtained  by  tossing  coins  or  throwing  dice 
are  said  to  depend  upon  C£  chance."  By  this  is  meant  that 
the  outcome  is  determined  by  a  large  number  of  similar, 
equal,  and  independent  factors.  The  height  from  which 
a  coin  or  die  is  thrown,  its  weight  and  size,  the  twist  of 
the  hand  employed,  and  many  similar  conditions  deter- 
mine which  particular  face  will  fall  uppermost.  Likewise, 
a  person's  height,  or  weight,  or  performance  on  an  intelli- 
gence test  can  be  regarded  as  depending  upon  a  variety 
of  independent  factors,  each  having  about  equal  influence 
upon  the  result.  Thus  it  has  been  suggested  that  the 
operation  of  chance  is  responsible  for  the  distribution  of 
human  traits  according  to  the  normal  frequency  curve. 

The  normal  curve  also  appears  in  a  different  situation 
as  the  curve  of  error.  When  repeated  measurements  are 
made,  the  results  will  not  be  identical  on  successive  occa- 
sions. These  errors  are  present  to  a  greater  or  lesser  degree 
in  all  types  of  measurement.  The  length  of  a  table  as 
measured  by  a  meterstick,  the  speed  of  a  simple  move- 


ment,  or  the  aesthetic  appeal  of  a  work  of  art  will  not 
remain  the  same  on  repeated  observations.  If  a  very 
large  number  of  observations  of  the  same  object  or  phe- 
nomenon are  made,  and  the  results  found  on  successive 
occasions  are  plotted  in  a  frequency  graph,  a  normal 
curve  will  be  obtained.  The  errors  of  observation  or 
measurement  which  produce  the  variation  are  themselves 
the  result  of  chance  factors,  and  hence  the  curve  of  error, 
like  the  distribution  curve,  will  coincide  with  the  normal 
probability  curve. 


The  implications  of  the  normal  distribution  curve  for  a 
psychology  of  individual  differences  can  be  realized  more 
vividly  by  contrasting  this  form  of  distribution  with 
other  possible  types.  The  distributions  chosen  in  partic- 
ular for  this  comparison  are  those  implied  by  certain 
common  theories  and  beliefs  in  regard  to  individual  dif- 
ferences. They  are  also  occasionally  found  with  actual 
test  results  because  of  the  use  of  faulty  techniques. 

A  skewed  distribution  is  one  in  which  the  peak  or 
"mode"  of  the  curve  is  displaced  to  either  side  of  the 
center.  Such  a  distribu- 
tion lacks  the  bilateral 
symmetry  of  the  normal 
curve.  In  Figure  4  will 
be  found  an  illustration  ; 
of  a  skewed  curve,  with  Scores 

piling  up  Of  SCOreS  at  the         FlG'  *  A  SKEWE»  DISTRIBUTION. 

upper  end  of  the  distribution.  Such  a  distribution  is 
implicit  in  the  popular  conception  of  many  character 
traits.  Thus  the  majority  of  people  are  considered  "hon- 
est" and  are  piled  up  at  one  extreme  of  the  scale;  from 


this  point,  the  number  of  cases  is  believed  to  decrease 
steadily  as  the  opposite  extreme  is  approached.  As  will 
be  shown  in  the  following  section,  this  type  of  distribution 
is  not  ordinarily  found  when  objective  measures  or  tests 
of  character  traits  are  used. 

A  skewed  distribution  may,  however,  be  obtained  with 
any  test  employed  on  a  group  to  which  it  is  not  suited. 
If  the  National  Intelligence  Test,  which  is  adapted  to 
grades  3  to  8,  were  administered  to  a  college  class,  the 
large  majority  of  subjects  would  score  very  near  the  max- 
imum, and  the  number  of  cases  would  decrease  rapidly 
towards  the  lower  scores.  Similarly,  if  one  of  the  many 
tests  constructed  for  use  on  college  freshmen  were  given 
to  elementary  schoolchildren,  there  would  be  a  marked 
piling  of  scores  near  the  zero  end  of  the  scale,  and  the 
distribution  would  be  equally  asymmetrical.  Obviously 
these  data  could  not  be  taken  to  mean  that  intelligence 
is  not  normally  distributed  among  schoolchildren  or  col- 
lege students.  Such  skewed  distributions  result  from  the 
fact  that  the  difficulty  range  of  the  test  does  not  extend 
far  enough  in  the  upper  or  lower  direction.  In  the  one 
case,  all  of  those  subjects  who  have  more  than  a  certain 
minimum  of  the  ability  tested  will  make  a  perfect  or  nearly 
perfect  score,  whereas  if  the  test  had  included  more 
difficult  items,  these  subjects  would  have  scattered  over 
a  wide  range.  The  same  holds  true  for  zero  or  very  low 
scores  when  the  test  is  too  difficult  for  the  group.  In 
choosing  a  test  for  a  given  group,  care  must  be  taken  to 
insure  that  the  subjects  have  sufficient  leeway  at  both 
ends  of  the  scale.  The  highest  and  lowest  scores  obtained 
should  be  a  considerable  distance  from  zero  and  perfect 
scores,  respectively. 

Skewness  may  also  result  from  the  inclusion  within  a 
single  distribution  of  two  normally  distributed  groups 


which  differ  pronouncedly  in  both  average  and  variability. 
This  effect  is  illustrated  in  Figure  5.     On  Graph  A  are 

A.  Two  groups  Plotted  Separately 

B,  Two  Groups  Combined 


given  the  separate  distribution  curves  of  the  two  groups, 
one  of  which  has  a  lower  average  as  well  as  a  narrower 
scatter  of  scores  than  the  other.  Graph  B  shows  the 
definitely  skewed  curve  which  is  obtained  when  both 
groups  are  combined  and  plotted  as  one  distribution. 

A  type  of  distribution  not  so  frequently  found  as  the 
skewed  curve  but  nevertheless  assumed  in  certain  common 
practices  is  the  rectangular  distribution  illustrated  in  Fig- 
ure 6.  If  individual  differences  were  distributed  in  this 
manner,  it  would  mean  that 
there  were  as  many  geniuses 
and  idiots  as  mediocre  people, 
as  many  men  whose  height  is 
6  feet  6  inches  as  those  whose 
height  is  5  feet  8  inches.  It  is 
interesting  to  speculate  on  the  effect  which  such  a  sit- 
uation would  have  on  our  sense  of  values.  Our  thinking  is  so 
permeated  with  the  knowledge  that  extreme  degrees  of  a 



trait  are  relatively  infrequent,  that  it  is  difficult  even  to 
conceive  of  a  world  in  which  extremity  did  not  imply  rarity. 

In  spite  of  this  fact,  the  assumption  of  a  rectangular 
distribution  is  sometimes  made  in  the  interpretation  of 
percentile  scores.  In  the  percentile  system  of  scoring, 
the  subject's  standing  on  any  test  is  expressed  in  terms 
of  the  percentage  of  people  in  a  given  group  whose  scores 
he  equals  or  excels.  If  an  individual,  for  example,  answers 
ten  questions  correctly  on  a  test,  and  if  50%  of  the  group 
complete  ten  questions  or  less  on  the  same  test,  then 
the  individual  falls  just  midway  in  the  group  and  is 
given  a  percentile  rating  of  50.  When  comparing  indi- 
viduals who  receive,  let  us  say,  percentile  scores  of  90, 
80,  60,  and  50,  we  must  bear  in  mind  that  the  difference 
in  ability  between  the  first  two  cases  is  greater  than  that 
between  the  last  two,  although  in  both  pairs  the  difference 
is  10  percentile  points.  In  order  to  include  10%  of  the 
cases  between  the  9Oth  and  Both  persons,  we  must  cover 
a  much  longer  distance  on  the  base  line  of  the  normal 
curve  than  is  necessary  in  going  from  the  5oth  to  the 
6oth  percentiles.  This  results  from  the  greater  clustering 
of  individuals  near  the  center  of  the  curve,  and  the 
relatively  small  number  of  cases  at  the  extremes.  Only 
if  the  distribution  were  rectangular  would  successive  per- 
centile scores  represent  equal  units  of  ability.  This  does 
not  mean  that  percentile  scores  are  of  no  value.  Like 
" mental  ages,"  they  furnish  a  simple  and  vivid  means 
of  expressing  the  subject's  standing  on  a  test.  Such  devices 
do  not,  however,  furnish  an  equal  unit  scale  of  ability. 

Lastly,  special  mention  should  be  made  of  the  multi- 
modal  distribution  because  of  the  prominent  part  it  plays 
in  current  "type  theories."  A  multi-modal  curve  is  one 
having  more  than  one  mode  or  peak.  Instead  of  a  single 
clustering  of  individuals  in  the  center  as  in  the  normal 


curve,  or  at  either  extreme  as  in  a  skewed  curve,  the 
clustering  occurs  at  several  points.  The  peaks  may  be 
equally  large,  or  there  may  be  a  major  peak  and  one  or 
more  minor  ones.  The  most  popular  variety  seems  to  be 
the  bi-modal  curve,  with  two  approximately  equal  peaks. 
All  of  the  common  schemes  of  classification  which  place 
individuals  into  distinct  categories  presuppose  some  form 
of  multi-modal  distribution.  The  division  of  men  into 
the  genius,  the  normal,  and  the  feebleminded,  the  sane 
and  the  insane,  the  sociable  and  the  unsociable,  all  rest 
upon  a  tacit  assumption  that  "most  people"  can  be 
classified  clearly  into  one  of  these  groups,  with  possibly 
a  few  intermediate  doubtful  cases.  It  is  interesting  to 
note  that  these  distinctions  are  much  less  common  in 
the  realm  of  physical  traits,  where  continuity  of  variation 
is  more  apparent  to  the  naked  eye. 

A  multi-modal  curve  can  be  obtained  in  any  trait 
through  the  arbitrary  selection  of  cases.  If  the  sampling 
tested  is  not  chosen  at  random  from  the  general  popula- 
tion, but  consists  of  individuals  selected  from  widely 
differing  levels  and  arbitrarily  combined  into  a  single 
group,  a  multi-modal  distribution  will  result.  A  group 
consisting  of  five-year-olds  and  ten-year-olds,  for  exam- 
ple, would  present  a  definitely  bi-modal  distribution  in 
intelligence  test  scores,  as  well  as  in  height,  weight,  and 
many  other  characteristics.  Were  the  intervening  age 
groups  from  six  to  nine  to  be  included  in  this  sampling, 
the  distribution  would  take  on  the  appearance  of  the 
normal  bell-shaped  curve. 

The  production  of  a  bi-modal  distribution  by  combining 
two  curves  of  widely  separated  groups  is  illustrated  in 
Figure  7.  It  will  be  noted  that  the  overlapping  between 
the  two  groups  is  very  slight.  When  the  overlapping  is 
large,  as  in  the  case  of  adjacent  age  groups,  the  resulting 


combined  curve  will  be  normal  and  uni-modal.  An  ex- 
ample of  a  bi-modal  curve  plotted  with  actual  scores  is 
presented  in  Figure  8.  The  two  distributions  which  are 
combined  in  this  curve  consist  of  the  Army  Alpha  scores 
obtained  by  two  groups  in  the  United  States  army.  The 
lower  group  includes  2773  native-born  white  soldiers  who 
had  reached  no  higher  than  the  fourth  elementary  grade 

A.  Two  Groups  Plotted  Separately 

B.  Two  Groups  Combined 


when  they  left  school;  the  upper  group  consists  of  3954  of 
ficers  who  had  had  four  years  of  college  work.  The 
combined  curve  exhibits  the  definite  bi-modality  which 
would  be  expected. 

We  have  seen  that  various  types  of  distributions  other 
than  the  normal  curve  are  frequently  implied  in  current 
popular  conceptions,  and  are  occasionally  found  because 
of  faulty  techniques  or  inadequate  data.  It  now  remains 
to  examine  some  typical  findings  on  trait  distributions 
obtained  under  conditions  relatively  free  from  the  errors 
described.  A  few  last  words  of  caution  should,  however, 
be  added.  In  the  first  place,  an  unlimited  number  of 
irregularities  and  variations  in  distribution  curves  may 
occur  through  the  use  of  small  groups.  Curves  plotted 


with  a  small  number  of  cases  present  an  uneven,  jagged 
appearance  since  individual  exceptions  loom  relatively 
large.  For  this  reason,  only  curves  obtained  on  fairly 
large  groups  will  be  considered  in  the  following  section. 

A  second  point  to  bear  in  mind  relates  to  the  concept 
of  a  mathematically  perfect  normal  curve  as  contrasted 

0-  10-  20-  30-  40-  50-  60-  70-  80-  90- 100-110-120-130-140-150-160-170-180-190-200. 
9  19  29  39  49  59  69  79  89  99  109  119  129  139  149  159  169  179  189  199  209 

Alpha  Scores 

AND  3954  OFFICERS  WITH  4  YEARS  OF  COLLEGE.  (Data  from  Yerkes,  20,  pp.  773, 

to  the  more  or  less  rough  approximations  obtained  with 
actual  data.  In  several  instances,  especially  when  the 
groups  were  very  large,  the  characteristics  of  the  distribu- 
tion curves  were  investigated  mathematically  and  found 
to  fall  within  the  expected  limits  of  normality.  In  many 
other  instances,  however,  where  such  mathematical  tests 
have  not  been  employed,  or  where  the  curve  has  been 
found  to  deviate  significantly  in  some  one  respect  from 


the  mathematically  normal  curve,  we  still  speak  of  "nor- 
mal distribution."  In  these  cases,  what  is  meant  is  that 
the  distribution  resembles  the  normal  curve  more  closely 
than  it  does,  let  us  say,  a  skewed,  rectangular,  or  multi- 
modal  curve.  It  exhibits  the  general  characteristics  of 
clustering  near  the  center  with  a  gradual  and  continuous 
sloping  toward  the  extremes.  For  a  general  picture  of 
the  distribution  of  a  trait,  such  knowledge  is  sufficient. 
Only  when  certain  statistical  techniques  which  assume  a 
normal  curve  are  to  be  applied  to  the  data,  is  it  necessary 
to  stress  the  requirement  of  mathematically  established 


In  Figures  9  to  22  will  be  found  examples  of  distribution 
curves  obtained  for  a  wide  variety  of  traits.  Only  distribu- 
tions based  on  large  samplings  have  been  included.  The 
two  smallest  groups,  comprising  200  and  400  cases,  re- 
spectively, are  taken  from  the  field  of  personality  testing. 
In  this  there  is  little  from  which  to  choose,  since  unfor- 
tunately very  few  frequency  tables  for  personality  tests 
are  available. 

An  example  of  the  distribution  of  a  purely  structural 
trait  is  furnished  in  Figure  9,  which  shows  the  height  in 
inches  of  6194  English-born  men.  It  will  be  seen  that  the 
graph  approximates  the  mathematical  normal  curve  to  a 
remarkably  close  degree.  Figure  10  presents  the  fre- 
quency curve  of  a  more  functional,  physiological  trait, 
vital  capacity.  This  is  the  total  volume  of  air,  measured 
in  cubic  centimeters,  that  can  be  expelled  from  the  lungs 
after  a  maximal  inspiration.  The  measurements  from 
which  the  curve  is  plotted  were  made  on  1633  male  stu- 
dents at  the  University  of  Minnesota.  The  general 
correspondence  to  the  normal  curve  is  again  apparent. 





i  600 


58   60   62   64   66   68   70   72   74   76   78   80 

Height  in  Inches 

(After  Yule,  21,  p.  89.) 



2  160 



1100   1700    2300    2900,   3500    4100   4700    5300    5900    6500    7100 

Vital  Capacity  in  Cubic  Centimeters 

(After  Jackson,  5,  p.  94.) 


The  two  graphs  reproduced  in  Figures  II  and  12  repre- 
sent the  distribution  of  performance  on  relatively  simple 
sensori-motor  and  mental  tests.  Reference  may  also  be 
made  in  this  connection  to  the  data  reported  previously 
in  Table  I  and  Figure  I.  All  three  sets  of  measures  were 

60-  70-  80-  90-  100- 110-  120-  130-  140-  150- 160-  170-  180-  190-  200- 
69  79  89  99  109  119  129  139  149  159  169  179  189  199  209 


STUDENTS.   (After  Anastasi,  2,  p.  32.) 

obtained  on  the  same  group  of  1000  college  students. 
The  tests  whose  distributions  have  been  reproduced  in- 
clude cancellation,  Pyle  symbol-digit,  and  a  nonsense- 
syllable  vocabulary  test.  In  the  first,  the  score  is  the  total 
number  of  A's  in  a  page  of  pied  type  cancelled  in  one 
minute.  This  is  generally  regarded  as  a  simple  test  of 
attention  and  perception,  although  speed  and  control  of 
movement  are  also  involved.  The  symbol-digit  test  is  a 
simple  learning  test  of  the  code  substitution  variety. 
The  vocabulary  test  is  a  more  difficult  learning  test,  also 


employing  a  code,  which  in  this  case  consists  of  paired 
nonsense  syllables.     The  distributions  of  all  three  tests 

10-  20-  30-  40-  50-  60-  70-  80-  90-  100-  110-  120-  130-  140- 
19  29  39  49  59  69  79  89  99  109  119  129  139  149 


(After  Anastasi,  2,  p.  34.) 

fall  within  the  expected  values  of  the  theoretical  normal 

Typical  results  obtained  with  intelligence  tests  on  large 
samplings  are  presented  in  Figures  13  to  16.  Figure  13 
gives  the  intelligence  ratings  on  the  "combined  scale" 
obtained  by  93,965  soldiers  in  the  United  States  army 
during  the  World  War.  This  group  constituted  a  random 
sampling  of  the  white  draft,  both  native  and  foreign- 
born.  The  "combined  scale"  ratings  are  based  on  Army 
Alpha,  Army  Beta,  Stanford-Binet,  and  performance  scale 

1  Mathematical  tests  of  normality  were  applied  to  these  curves  (cf.  Ana- 
stasi, 2). 





•      =» 

i  7 

§    6 


°-  5 

0  5  10  15 

Combined  Scale  Scores 

(N  =  93,965).   (Data  from  Yerkes,  20,  p.  654.) 



w  OK 
</>  ^0 


"a  20 




I  10 

56-     66-    76-     86-    96-    106-   116-    126-    136- 
65      75     85      95     105     115    125     135     145 

Standford-Binet  I.Q, 

FIG.  14.  I.Q/s  OF  905  UNSELECTED  CHILDREN,  5-14  YEARS  OF  AGE, 
(After  Terman,  14,  p.  66.) 


examinations,  the  scores  on  all  tests  having  been  reduced 
to  comparable  terms  and  expressed  on  a  single  scale 
ranging  from  o  to  25  points.  Since  all  the  recruits  were 
unable  to  take  the  same  test,  it  was  necessary  to  combine 
scores  in  this  manner. 

In  Figure  14  will  be  found  the  distribution  of  Stanford- 
Binet  I.Q.'s  of  905  unselected  schoolchildren,  ranging  in 
age  from  5  to  14.  The  percentages  of  children  whose 
I.Q.'s  fell  within  each  class-interval  are  given  below  the 
graph.  Figure  15  shows  the  percentage  distribution  of 
scores  made  by  5952  sixth- 
grade  schoolchildren  on 
the  Otis  Advanced  Exam- 
ination. This  is  a  widely 
used  group  test  of  general 
intelligence.  Finally,  in 
Figure  16  can  be  seen  a 
composite  curve  for  college 
freshmen,  obtained  by  ^ 

,  .    .  -        FIG.  15.    PERCENTAGE  DISTRIBUTION 

Combining    the    results    OI    OF  SIXTH  GRADE  SCORES  IN  THE  OTIS 

eleven     different     intelli- 

gence    examinations     ad- 

ministered to  groups  of  623  to  5495  freshmen  in  various 

American  colleges.  The  theoretical  normal  curve  is  shown 

in  broken  line  on  the  same  graph. 

In  the  measurement  of  personality  and  character  traits, 
testing  techniques  are  still  in  a  relatively  crude  and  un- 
developed stage.  Many  sources  of  error  remain,  so  that  one 
should  scarcely  expect  to  find  perfect  specimens  of  the  nor- 
mal distribution  curve.  Despite  a  more  jagged  appearance 
and  many  minor  irregularities,  however,  the  available 
distribution  curves  exhibit  quite  generally  the  fundamen- 
tal characteristics  of  the  normal  curve.  Inspection  of 
Figures  17  to  22  will  make  this  apparent. 


THEORETICAL  NORMAL  CURVE.  (After  Thorndike,  15,  p.  545.) 

-41-  -36-  -31-  -26-  -21-  -16-  -11-  -6-    -1-     0-  +5-  +10-  +15-  +20- 
-45  -40  -35  -30  -25  -20   -15  -10    -5    +4    +9    +14  +19  +24 

Extroversion  -« Scores *~  Introversion 

200  COLLEGE  STUDENTS.  (Data  from  Heidbreder,  9,  p,  124.) 








1 50 





+55-  +45-  +35-  +25-  +15-    +5-    -5-   -15-  -25-  -35-  -45- 
+46   +36    +26    +16     +6    -4    -14    -24   -34   -44    -55 


ASCENDANCE-SUBMISSION  TEST.    (After  Allport,  i,  p.  129.) 

Figure  17  gives  the  distribution  of  total  introversion- 
extroversion  scores  on  a  self-rating  questionnaire  admin- 
istered to  200  college 
students.  The  posi- 
tive scores  correspond 
to  the  introvert  end 
of  the  scale,  the  nega- 
tive scores  to  the  ex- 
trovert end.  It  will  be 
readily  seen  that  in- 
dividuals do  not  clus- 
ter at  opposite  ends 
of  the  scale,  as  a  clear- 
cut  division  into  in- 
troverts and  extro- 





\  150 



0    .1     .2 

.8    .9   .10 

3    .4     .5    .6     .7 
Cheating  Ratio 


^  ij      •     ^u,     TIOS"  OF  2443  SCHOOLCHILDREN.    (Data  from 

VertS     WOUld     imply.    Hartshorne  Ind  May,  6,  p.  220.) 

The  greatest  cluster- 
ing occurs  in  the  center,  with  a  gradual  dropping  off  as  the 
extremes  are  approached.  Figure  18,  giving  the  distribution 


of  400  male  college  students  on  a  test  of  ascendance- 
submission,  exhibits  the  same  general  form. 

Figures  19  to  22  are  plotted  with  data  taken  from  the 
studies  of  May  and  Hartshorne  (6,  7,  8)  on  the  measure- 
ment of  character  in  schoolchildren.    Figure  19  gives  the 
distribution  of  "cheating  ratios5'  for  2443  children.    The 
no  cheating  ratio  indi- 

A  cates    the    number 

100   "  /   \  r     •  11-11 

/    \  ot  times  each  child 

90  -  /        \ 

cheated  relative  to 
the  number  of  op- 


!  70 
•  60 

JS  50 




portunities  offered. 
The  obtained  curve 
does  not  admit  of 
a  clear-cut  division 
into  "honest"  and 
"dishonest,"  or 
those  who  cheat 

44.  16-  46-  76.106-136-166-196-226-256.286-316.    ™d    ^O8e  Wh°   d° 
15    45    75   105  135  165  195  225  255  285  315  345    not.   A  slight  skew- 

FIG.  20.  DISTRIBUTION  OFepERsiSTENCE  SCORES  n^SS  1S  ex^lblted, 
AMONG  656  SCHOOLCHILDREN.  (Unpubl.  data  from  with  a  tendency  for 
investigation  of  Hartshorne,  May,  and  Mailer,  7.)  ,  •-, 

"    scores  to  pile  up  at 

the  "honest"  .end,  but  this  may  be  caused  by  a  limitation  in 
the  scale.  The  tests  probably  presented  an  insufficient  num- 
ber of  situations  in  which  cheating  was  made  very  easy  or 
involved  a  relatively  minor  "moral  issue."  This  would  cut 
the  scale  short  at  the  lower  end  and  produce  an  excess  of  zero 
or  very  low  cheating  scores.  The  distribution  of  combined 
scores  on  several  tests  of  service  or  cooperativeness,  pre- 
sented in  Figure  20,  exhibits  a  very  close  resemblance  to 
the  normal  curve.  The  same  general  features  also  char- 
acterize the  distributions  in  Figures  21  and  22.  The 
former  gives  the  combined  scores  on  several  persistence 


60-  65-  70-  75-  80-  85-  90-  95- 100-105- 110-115- 120- 125- 130-135- 140- 145-150- 155-160- 
64  69  74  79  84  89  94  99  104  109  114  119  124  129  134  139  144  149  154  159  164 


(UnpubL  data  from  investigation  of  Hartshorne,  May,  and  Mailer,  7.) 

tests;  the  latter  is  based  on  the  results  of  a  question- 
naire of  moral  knowledge,  designed  to  measure  "good 


One  is  tempted  to  compare  the  distributions  of  different 
traits  in  the  effort  to  discover  the  relative  variability  of 
such  traits.  Do  individuals  differ  more  in  physical  or  in 
mental  traits?  Are  they  more  alike  in  intellectual  or  in 
emotional  characteristics  ?  These  and  many  similar  ques- 
tions have  been  raised  repeatedly  and  answers  have  oc- 
casionally been  offered.1  It  is  probably  correct  to  state 

•  l  Cf.,  for  example,  the  interesting  although  rather  futile  discussion  by  Wech- 
sler  (19). 


as  a  general  principle  that  individual  differences  will  be 
larger  in  the  more  complex  than  in  the  simpler  traits. 
Any  characteristic  which  depends  upon  the  simultaneous 
variation  of  a  large  number  of  factors  will  exhibit  more 






$  80 


j§  60 
£  50 





i     a     \     i     i     \     till 



10-  15-  20-  25-  30-  35-  40-  45-  50-  55-  60-  65-  70-  75-  80- 
14  19  24  29  34  39  44  49  54  59  64  69  74  79  84 


(Unpubl.  data  from  Hartshorne,  May,  and  Shuttleworth,  8.) 

marked  differences  than  one  which  is  determined  by  rela- 
tively few  factors.  An  illustration  from  coin  tossing  will 
again  prove  serviceable.  If  two  coins  are  employed,  the 
number  of  possible  combinations  which  may  result  is 
only  four;  if,  however,  the  number  of  coins  is  increased 
to  ten,  the  possible  variations,  or  patterns  of  head-and- 
tail  combinations,  total  1024.  A  complex  trait  is  one 
which  is  determined  by  a  large  number  of  factors  or  con- 
ditions, and  hence  it  will  be  expected  to  exhibit  a  greater 
range  of  variation. 


Apart  from  this  rather  obvious  generalization,  little  can 
be  said  about  the  extent  of  individual  variation  in  differ- 
ent traits.  Upon  close  analysis,  in  fact,  the  question  of  the 
extent  of  variability  itself  appears  to  be  ambiguous  and 
quite  meaningless.  The  first  problem  which  confronts 
one  when  trying  to  compare  human  variability  in  sep- 
arate traits  is  that  of  the  measuring  rod  employed  for  the 
different  traits,  or  the  units  in  which  the  measurements 
are  reported.  That  the  particular  scale  employed  affects 
the  amount  of  variability  found  is  easily  demonstrated. 
If  the  height  of  buildings  in  one  city  is  measured  in  feet 
and  in  another  city  in  yards,  the  buildings  in  the  former 
city  will  seem  to  vary  among  themselves  three  times  as 
much  as  in  the  latter,  even  though  the  actual  range  in 
height  may  be  identical  in  the  two  cities.  Fortunately, 
feet  can  be  translated  into  yards  and  vice  versa.  But 
this  cannot  be  done  with  the  units  of  psychological  tests. 
The  number  of  problems  correctly  solved  on  an  arithmetic 
test  cannot  be  transmuted  into  the  same  kind  of  units  as 
neurotic  symptoms  checked  on  an  emotional  adjustment 
questionnaire.  The  only,  solution  offered  for  this  difficulty 
is  the  use  of  measures  of  relative  variability. 

All  indices  of  relative  varUbility.^ig^raiiS8*    One  such 
measure,  the  coeffici^^^oj^variation^  is JOUE^^ 
the  standard  deviation  1  by  the  ayej;a^rg|lth^^ 
Tims  variability  is  expressed  in  relation  to  the  average, 
the  difference  in  units  from  one  test  to  the  other  being 
automatically  ruled  out.   For  the  same  purpose,  thojj-lia 
between  the  highest  and  lowest  scores  or  the  tenthJhigiLjest 
tand  tenth  lowest,  or, any,  other  similar^coi^^ 

1  The  standard  deviation  is  a  statistical  measure  of  the  extent  of  variability 
within  a  group.  It  is  found  by  subtracting  each  score  from  the  average,  squaring 
the  differences,  and  then  finding  the  square  root  "of  the  average  of  these  squares. 


frequently  cpmputed.  Although  in  current  use,  all  such 
measures  are  open  to  a  serious  objection.  The  difficulty 
arises  from  the  fact  that  psychological  scales  do  not 
measure  the  individual  from  a  true  zero  of  ability  as  a 
base.  Thus  a  zero  score  on  Army  Alpha  does  not  mean 
zero  intelligence,  as  was  demonstrated  when  soldiers  who 
had  failed  to  make  any  score  on  the  Alpha  were  given  the 
Beta  or  the  Stanford-Binet  examinations.  The  Army 
Alpha  begins  at  an  arbitrary  level  of  ability  which  is 
higher  than  that  reached  by  some  individuals;  anyone 
falling  below  that  level  will  receive  a  zero  score. 

The  custom  of  measuring  from  "absolute  zero"  in  our 
physical  scales  is  so  general  that  it  is  difficult  to  conceive 
of  the  effects  of  using  a  scale  that  begins  at  an  arbitrary 
zero  point.  Think  of  a  measuring  stick  on  which  height 
is  measured,  not  from  absolute  zero  or  no  height  at  all, 
but  from  some  arbitrary  point  such  as  two  feet.  The  fol- 
lowing diagram  illustrates  the  situation.  Any  object 


4 -FOOT  BOY  6-Foi)T  MAN 

two  feet  or  less  in  height  would  register  zero  on  this  scale. 
If  such  a  scale  were  to  be  employed  only  to  measure  the 
heights  of  individuals  over  five  years  of  age,  the  arbitrary 
limit  would  perhaps  not  appear  so  absurd,  since  no  one 
would  be  under  two  feet  tall.  This  is  in  fact  what  has 
occurred  in  the  construction  of  psychological  tests.  Since 
the  Army  Alpha,  for  example,  was  designed  for  adult 
men,  it  would  have  been  wasteful  and,  from  a  practical 
standpoint  impossible,  to  extend  it  down  to  the  intel- 
lectual level  of  a  newborn  child.  To  return  to  our  yard- 
stick with  an  arbitrary  zero  point  at  two  feet,  let  us 
suppose  that  it  has  been  used  to  measure  the  heights  of  a 
six-foot  man  and  a  four-foot  boy.  The  man  will  measure 


four  feet  and  the  boy  two  feet,  as  has  been  indicated  on 
the  diagram.  For  many  purposes,  no  error  has  been 
introduced  in  the  data  by  the  use  of  the  artificial  zero 
point.  On  any  scale,  the  man  is  two  feet  taller  than  the 
boy.  If,  however,  we  express  their  respective  heights  as  a 
ratio,  we  reach  the  conclusion  that  the  man  is  twice  as 
tall  as  the  boy  (-f).  This  is  not  true  of  their  actual  heights 
from  absolute  zero,  the  man  being  only  i|  times  as  tall 
as  the  boy  (f).  The  subtraction  of  a  constant,  two  feet, 
from  both  heights  has  yielded  a  false  ratio. 

Such  is  the  effect  of  aa  .arbitrary  zero  point  on  any 
value  which  involves  the  division  of  one  measure  by  an- 
other. For  this  reason,  ratio  or  other  relative  measures 
cannot  be  employed  with  the  large  majority  of  psycho- 
logical tests  which  are  not  scaled  from  absolute  zero.1 
Such  measures  would  hold  true  only  for  the  specific  tests 
in  the  form  in  which  they  were  employed;  the  addition 
or  removal  of  a  few  easy  items  at  the  lower  end  of  the 
scale  would  completely  alter  the  relative  variabilities. 
Obviously  the  values  thus  computed  could  not  be  regarded 
as  very  meaningful.  We  arrive  at  the  collusion,  then,, 
that  with  available  psychological  tests  it  is  imposjible^to 
compare  variability  from  one  trait  to  another. 

Other  difficulties  also  appear  as  the  problem  is  inspected 
more  closely.  I^gjthe^ 

it^refer^  to  the  whole  human  race?  Which  individuals, 
if  any,  shall  be  omitted  m' order  to  arrive  at  an  estimate 
of  human  variability?  Shall  those  who  are  regarded  as 
definitely  pathological  and  represent  very  extreme  devia- 
tions be  excluded?  If  so,  where  should  the  line  be  drawn 
between  a  typical  human  group  and  an  abnormal  deviant? 

1  The  only  important  exception  to  date  is  the  CAVD  Intelligence  Examina- 
tion, prepared  by  the  Institute  of  Educational  Research  at  Teachers'  College, 
Columbia  University  (cf.  Thorndike,  15). 


It  seems  reasonable,  for  example,  to  exclude  from  an 
estimate  of  the  range  of  human  variation  in  speed  of 
movement  one  who  has  suffered  an  injury  which  renders 
his  movements  slow  and  halting.  It  is  but  a  short  step 
from  this  procedure  to  that  which  would  exclude  those 
incapacitated  through  disease.  How,  then,  would  this 
criterion  operate  in  the  case  of  a  feebleminded  person 
in  whom  no  physical  defect  can  be  discovered  ?  How  far 
shall; this  process  of  eliminating  extreme  cases  be  carried? 

A  further  question  relates  to  the  factors  which  are  to  be 
held  constant  in  measuring  the  variability  of  any  one 
trait.  How  homogeneous  should  the  group  be?  The 
inclusion  of  children  of  different  age  levels  would  cer- 
tainly, increase  the  extent  of  variation  in  most  traits.  If 
only  the  range  of  individual  differences  within  a  fairly 
homogeneous  population  is  desired,  the  difficulty  of  de- 
fining the  required  degree  of  homogeneity  is  encountered. 
Many  traits  are  influenced  by  the  social  and  economic 
level  in  which  the  individual  finds  himself.  Should  condi- 
tions of  this  sort  also  be  held  constant?  Should  differences 
in/;  speed,  of  performance  be  ruled  out  when  determining 
variability  in  "intelligence"?  Such  questions  could  be 
raised  ad  inftnitum.  unless  an  arbitrary  limit  is  set  up  and 
adhered  to  consistently  for  the  purposes  of  some  one 
particular  investigation. 

We  may  conclude  from  this  analysis  that  the  question . 
of,  the  extent  of  individual 'differences  in  different  traits 
cannot  be  answered  unless  put  in  very  specific  terms. 
The  population  must  be  defined  in  detail  within  each 
investigation  and  the  nature  of  the  trait  measured  must 
be  made  clear,  especially  by  indicating  which  conditions 
are  to  be  held  constant  and  which  will  be  allowed  to  vary. 
Obviously  all  conditions  whjiih  affect  a  given  trait  cannot 
be  held  constant;  otherwise  variation  would  disappear. 


It  should  be  added  that  at  the -present  stage  in  the  .de- 
velopment of  mental  testing,  owing  to  the  use  of  incom- 
parable;, units  and  arbitrary  zero  points,  the  question 
cannot  be  answered  at  all,  in  any  form. 


Individual  differences  ,are  not  to  be  regarded  as '.char- 
acteristically human.  Variation  is  a  universal  phenomenon 
throughout  the  organic  scale.  "All  cats  look  grey  at 
night/'  but  upon  closer  inspection  each  becomes  an  indi- 
vidual in  his  own  right.  Cursory  or  inadequate  observa*- 
tion  often  creates  an  impression  of  similarity  :brrey,eh 
identity  among  members  of  a  group  while  the  differences 
pass  unnoticed.  For  this  reason,  only  the  very  exceptional 
deviations  among  animals  have  attracted  attention  in  the 
past,  all  other  members  of  the  species  having  been  im- 
plicitly relegated  to  the  limbo  of  "normality." 

Several  cases  of  exceptionally  "gifted"  animals <x  have 
been  described  by  their  trainers  or  by  observers,  the 
remarkable  feats  of  the  animals  having  aroused  the  wonder 
"and  admiration  of  spectators.  Among  the  most  famous 
examples  is  Clever  Hans,  a  stallion  purchased  in  1900 
by  a  Mr.  Van  Osten  of  Berlin  and  subsequently  trained 
by  him.  The  horse  was  first  taught  a  conventional  alphas- 
bet  in  which  each  letter  was  represented  by  :a  fcertain 
combination  of  taps  with  the  forefoot.-  Digits  -were;  iiiflir 
cated  by  the  appropriate  number  of  taps;  By  this  sys,tein| 
the  horse  learned  to  "count "--objects  presented  *  to ";Mni 
and  also  to  perform  all  forms  of -simple  arithmetic  opasa^- 
lions:  "He.  .could  handle  fractions,  first  changing:  them 
into  decimals;  "He  was  able  to  give  the  correct  answer 
to  such  a  problem  as  the  following:  "I  "have  a  number  in 
:mind;  1  subtract  9  and  have  3  as  a  remainder;  what1  is 

i  For  a  fuller  discussion  sfee  Watson  (18,  Ch.  IX)  and  Tinklepaugh  (16). 


the  number?"  He  seemed  to  read  German  readily,  and 
if  presented  with  a  series  of  cards  containing  written 
words,  he  would  step  up  and  point  with  his  nose  to  any 
words  required  of  him.  He  answered  simple  questions 
put  to  him  orally,  tapping  out  each  letter  of  the  answer 
in  his  conventional  alphabet.  He  could  give  the  date  of 
any  day  one  might  mention,  would  tell  time  to  the  min- 
ute, and  was  able  to  analyze  a  discordant  clang,  telling 
his  observers  which  note  should  be  changed. 

Most  of  these  feats  are  not,  to  be  sure,  as  remarkable 
as  they  appear  at  first  glance.  Thus,  it  was  found  that 
Clever  Hans  was  unable  to  respond  correctly  to  a  prob- 
lem if  no  one  present  knew  the  answer.  Likewise,  when 
the  observers  were  concealed,  the  horse  failed.  The  un- 
usual achievements  of  Clever  Hans  and  of  many  other 
performing  animals  result  not  from  an  understanding 
of  arithmetic  or  an  ability  to  read  but  from  an  exception- 
ally keen  observation  of  slight  cues  given  by  the  observers. 
The  trainer,  or  other  persons  present,  will  make  some 
slight  gesture,  such  as  lifting  the  head  a  few  millimeters, 
as  soon  as  the  animal  has  tapped  the  correct  number  of 
times.  Such  cues,  it  may  be  added,  are  usually  given  un- 
intentionally and  unconsciously.  They  may  be  too  slight 
to  attract  the  attention  of  spectators,  but  an  observant 
animal  will  learn  to  respond  to  them.  Although  destroy- 
ing some  of  the  glamor  which  such  feats  have  had  for  the 
public,  this  explanation  does  not  imply  that  the  task  of 
learning  to  observe  and  respond  to  the  proper  cues  is  an 
easy  one  which  any  animal  could  accomplish. 

There  remain,  furthermore,  the  cases  of  animals  who 
have  been  shown  genuinely  to  respond  to  a  wide  variety 
of  verbal  commands  in  the  absence  of  any  other  cues,  or 
who  have  learned  intricate  combinations  of  movements, 
or  have  in  many  other  ways  proved  their  ability  to  react 


to  very  complex  situations.  Performing  dogs,  such  as 
"Fellow"  who  could  respond  to  approximately  400  words 
and  execute  the  same  commands  even  when  worded  dif- 
ferently, have  been  repeatedly  exhibited.  Dogs  who  lead 
the  blind  show  a  remarkably  keen  adjustment  of  their 
responses  to  the  changing  demands  of  the  situation.  Chim- 
panzees have  been  taught  a  wide  variety  of  acts  such  as 
skating,  riding  a  bicycle,  eating  with  knife  and  fork, 
unlocking  doors.  The  performances  of  circus  animals, 
and  especially  "musical"  sea  lions,  are  well  known.  The 
observation  of  such  animals,  even  when  stripped  of  pop- 
ular overstatement,  still  yields  instances  of  marked  indi- 
vidual differences. 

Nor  is  the  evidence  for  individual  variation  among 
infrahuman  animals  confined  to  the  study  of  unusual 
cases.  Every  laboratory  investigation  employing  more 
than  one  subject  has  revealed  individual  differences.1 
Animal  psychologists  have  not  as  a  rule  been  concerned 
with  the  measurement  of  variability,  so  that  the  data  on 
this  problem  are  usually  mentioned  only  incidentally  and 
frequently  are  not  given  in  quantitative  form.  Whenever 
such  data  are  reported,  however,  the  range  of  perform- 
ance in  a  randomly  selected  group  is  surprisingly  large. 
Wide  individual  variation  has  been  found  in  every  phase 
of  behavior  investigated,  such  as  the  amount  of  general 
spontaneous  activity,  the  relative  strength  of  motives, 
speed  of  movement,  quickness  of  learning  simple  tasks, 
and  behavior  in  more  complex  problem-solving  situations. 
Some  typical  quantitative  results  on  learning  behavior 
have  been  brought  together  in  Table  II.  The  average, 
range,  and  standard  deviation  for  each  set  of  data  have 
been  given  whenever  available. 

1  Cf.,  for  example,  the  discussion  of  this  problem  from  various  angles  by 
Tryon  (17). 


;  first  set  of  data  are  taken  from  experiments  on 
conditioning.  Two  stimuli,  -such  as  a  flash  of  light  and  an 
electric  shock  to  the  foot,  are  presented  together.  After 
a  number  of  combined  repetitions  of  these  stimuli,  the 
withdrawal  response  becomes  conditioned  to  the  light, 
i.e-j  the  animal  will  withdraw  its  foot  upon  appearance 
of  the  light  alone,  without  the  presence  of  the  electric 
shock.  It  is  customary  in  such  an  experiment  to  refer 
to  the  original  stimulus  (in  this  case,  the  shock)  as  the 
conditioning  stimulus,  and  to  the  other  as  the  condi- 
tioned stimulus.  The  general  nature  of  the  condition- 
ing  and  conditioned  stimuli  employed  in  each  experi- 
ment has  been  indicated  in  Table  II,  together  with  the 
type  and  number  of  animals  investigated.  It  will  be 
noted  that  the  number  of  combined  repetitions  of  the 
two  stimuli  required  to  establish  the  conditioned  reaction 
differs  widely  from  individual  to  individual  within  each 

The-  second  set  of  data  are  taken  from  a  series  of  learning 
projects  conducted  at  the  Columbia  University  laboratory 
of  comparative  psychology.  Small  samplings  of  guinea 
pigs,  albino  rats,  common  short-haired  cats,  and  monkeys 
of.twb  species  Were  tested  with  the  same  type  of  "  problem 
box7'  in  which  a  series  of  steps  of  -•  increasing  complexity 
is>  presented  to  the  animal.  The  box  consists  essentially 
of  an  outer  and  an  inner  cage,  the  latter  containing  the 
incentive  which  the  animal  obtains  at  the  completion  of 
each  successful  trial.  In  the  outer  cage  are  three  plates 
to  be  depressed  in  a  given  order  by  the  animal  before  the 
door  to  the  incentive  compartment  is  opened.  In  Table  !!> 
only'  the  number  of  trials  required  to  learn  step  I  ,are 
reproduced,  since  this  is  the  only  step  learned  by  all>of 
the  groups.  The  problem  in  step  I  consists  simply  in 
stepping  on  the  first  plate  to  the  right  as  the  animal  en- 






Combinations for  Condi- 


No.  of 

















Light      - 


























PROBLEM  Box  f 

1  Organism 

No.  of 

Step  I 

Trials  to  Learn  Step  I 

Range  in 























46.69  J 






















Type  of  Maze 

No.  of 

/""./T  ..  -,_ 

No.  of  Trials  to,  Learn 


Range  " 

?•  s.p. 

Albino  rats 

8  cul-de-sac  ele- 

;' / 

1     i  :  i 

vated  skeleton 

-V*;  .,    <; 


1  86 



Albino,  rats 

Equal-unit  maze 




*  From  Razran  (13),  pp.  308-309. 
t  From  Fjeld  (4),  p.  528,  and  Koch  (11),  pp.  186,  208. 
J  Cf.  footnote  on  p.  62. 

§  From  Corey  (3),  p.  256,  and  Jackson  (10),  p.  27. 



ters.1  The  other  steps  involved  stepping  on  plates  I  and  2; 
I,  2,  and  3;  I,  2,  3,  and  back  to  2;  I,  2,  3,  2,  i;  and  so 
on  to  other  combinations.  Although  these  studies  were 
conducted  mainly  to  determine  the  highest  number  of 
steps  which  any  animal  within  a  given  species  could 
master,  the  data  yield  striking  evidence  of  individual 
differences  within  each  species.  Not  only  the  number  of 
trials  required  to  learn  each  step,  but  also  the  number 
of  steps  which  could  be  learned,  differed  from  individual 
to  individual.  Thus  in  the  group  of  guinea  pigs,  some 
were  unable  to  learn  even  step  I,  while  others  succeeded; 
among  the  rats,  some  learned  two  steps,  some  one,  and  a 
few  none;  among  the  cats,  the  range  is  from  3  to  7  steps; 
among  the  rhesus  monkeys  2  to  22,  and  among  the  cebus 
5*  to  15.  Thus  the  individual  variation  was  so  large  that 
an  individual  could  easily  be  found  in  a  "higher"  species 
who  was  unable  to  learn  as  much  as  a  given  individual 
in  a  "lower"  species. 

In  the  third  section  of  Table  II  are  presented  some  typ- 
ical data  on  maze  learning  among  albino  rats.  The 
individual  differences  are  again  marked,  as  is  indicated 
by  the  standard  deviations  of  the  number  of  trials  re- 
quired to  master  the  correct  path  in  each  maze.  Thus  it  is 
apparent  that  close  observation  and  measurement  of  ani- 
mal behavior  reveal  fully  as  wide  an  individual  variation 
as  the  studies  on  human  subjects. 

An  interesting  example  of  the  normal  distribution  curve 
in  a  functional  trait  in  animals  is  to  be  found  in  the  photo- 
graph and  accompanying  curve  reproduced  in  Figure  23, 
The  photograph  shows  horses  on  the  race  track  just 
before  the  finish.  The  relative  position  of  the  horses 

1  In  the  study  on  cats,  the  problem  set  in  step  I  was  simpler,  the  animal  being 
allowed  to  step  on  any  one  of  the  three  plates.  The  data  on  this  group  are  there- 
fore not  directly  comparable  to  those  on  the  other  species. 


,9608  .9650 .9686        .9753         .9811  .9847      .9875     .9903       .9922 

Quality  of  Performance  or  Racing  Capacity  Shown 

EPSOM  DOWNS,  JUNE  5,  1929.  (After  Laughlin,  12,  p.  215.) 

furnishes  a  vivid  demonstration  of  the  normal  distribution 
of  racing  performance.  A  few  are  in  the  lead,  an  equally 
small  number  lag  behind,  and  the  majority  are  scattered 
in  intermediate  positions.  The  graph  is  a  frequency  curve 
of  the  "racing  capacity"  of  the  same  horses,  computed 
by  a  standardized  formula. 



1.  Allport,   G.  W.      "A  Test  for   Ascendance-Submission," 
/.  Abn.  and  Soc.  PsychoL,  1928-29,  23,  118-136. 

2.  Anastasi,  A.     "Practice  and  Variability,"  PsychoL  Mon., 
i934?  45.  No.  5.  Pp.  55. 

3.  Corey,  S.  M.    "An  Experimental  Study  of  Retention  in  the 
White  Rat,"  /.  Exp.  PsychoL,  1931,  14,  252-259. 

4.  Fjeld,  H.  A.    "The  Limits  of  Learning  Ability  in  Rhesus 
Monkeys,"  Genet.  PsychoL  Mon.,  1934,  15,  369-537. 

5.  Harris,  J.  A.,  Jackson,  C.  M.,  Paterson,  D.  G.,  and  Scam- 
mon-,  R<  L.    The  Measurement  of  Man.   Minneapolis :  Univ. 
Minn.  Press,  i§3o.   Pp.  215. 

6.  Hartshorne,  H.,  and  May,  M*  A.   Studies  in  Deceit.   N.  Y.: 
Macmillan, '1928.  Pp.  306. 

7.  Hartshorne,  H.,  May,  M.  A.,  and  Mailer,  J.  B.    Studies  in 
Service  and  Self-Control.   N.  Y.:  Macmillan,  1929.    Pp.  559. 

8.  Hartshorne,  H.,  May,  M.  A.,   and  Shuttleworth,  F.  K. 
Studies  in  the  Organization  of  Character.   N.  Y. :  Macmillan, 
1930.   Pp.  503. 

9.  Heidbreder,  E.     "Measuring  Introversion  and  Extrover- 
sion," J.  Abn.  and  Soc.  PsychoL,  1926,  21,  120-134. 

10.  Jackson,  T.  A.   "General  Factors  in  Transfer  of  Training  in 
the  White  Rat,"  Genet.  PsychoL  Mon.,  1932,  n,  1-52. 

11.  Koch,  A.  M.     "The  Limits  of  Learning  Ability  in  Cebus 
Monkeys,"  Genet.  PsychoL  Mon.,  1935,  I7?  I^S~~::i34- 

12.  Laughlm,  H.  H.    "Racing  Capacity  In  the  Thoroughbred 
Horse,"  Sci.  Monthly,  1934,  38,  210-222;  310-321. 

13.  Razran,  G.  H.  S.     "Conditioned  Responses  in  Animals 
Other  than  Dogs,"  PsychoL  Bull.,  1933,  30,  261-324. 

14.  Terman,  L.  M.     The  Measurement  of  Intelligence.    N.  Y.: 
Houghton  Mifflin,  1916.    Pp.  362. 

15.  Thorndike,  E.  L.,  Bregman,  E.  0.,  Cobb,  M.  V.,  and  Wood- 
yard,  Ella.    The  Measurement  of  Intelligence.   N.  Y.:  Teach- 
ers College,  Columbia  Univ.,  Bur.  Pub.,  1926.  Pp.  616. 

16.  Tinklepaugh,  O.  L.    "'Gifted'  Animals."   Ch.  XV  in  Com- 


parative  Psychology •,  ed.  by  Moss,  F.  A.    N.  Y. :  Prentice- 
Hall,  1934.   Pp.  529. 

17.  Tryon,  R.  C.  "Individual  Differences."  Ch.  XIII  in  Compar- 
ative Psychology ,  ed.  by  Moss,  F,  A.    N.  Y. :  Prentice-Hall, 
1934.  Pp.  529. 

1 8.  Watson,  J.  B.     Behavior:  an  Introduction  to  Comparative 
Psychology.   N.  Y.:  Holt,  1923.   Pp.  439. 

19.  Wechsler?  David.    The  Range  of  Human  Capacities.    Balti- 
more: Williams  and  Wilkins,  1935.  Pp.  159. 

20.  Yerkes,  R.  M.,  ed.  "Psychological  Examining  in  the  United 
States  Army,"  Memoirs  Nat.  Acad.  Sri.,  1921,  15.    Pp.  890. 

21.  Yule,  G.  U.     An  Introduction  to  the   Theory  of  Statistics, 
London:  Griffin,  1919.    Pp.  422. 


Why  do  individuals  differ  from  one  another?  What 
are  the  factors  which  produce  variation?  These  questions 
have  stimulated  intense  discussion  and  led  to  lively  con- 
troversy. In  addition  to  its  fundamental  theoretical  im- 
port, the  problem  of  the  causation  of  individual  differ- 
ences has  far-reaching  practical  significance  in  many  fields. 
Any  procedure  involving  the  control  of  human  develop- 
ment must  be  based  upon  an  understanding  of  the  factors 
which  influence  such  development.  All  educational  meth- 
ods make  some  assumption  regarding  the  causes  of  indi- 
vidual differences.  Is  the  main  function  of  education  to 
produce  certain  desirable  traits,  or  merely  to  offer  oppor- 
tunities for  the  development  of  the  child's  potentialities? 
Volumes  have  been  devoted  to  argumentative  and  fre- 
quently verbose  analyses  of  this  question.  The  empirical 
accumulation  of  facts  on  the  causes  of  individual  variation 
alone  can  furnish  a  conclusive  answer.  The  type  of  edu- 
cational activities,  vocations,  and  other  pursuits  tradi- 
tionally allotted  to  men  and  women  rests  upon  certain 
beliefs  regarding  the  cause  of  sex  differences  in  psycho- 
logical traits.  Relationships  among  racial  and  national 
groups,  and  the  attitudes  of  one  group  towards  another, 
are  built  up  on  the  basis  of  theories,  either  implicitly 
assumed  or  overtly  stated,  regarding  the  origin  of  racial 
and  national  characteristics.  Any  caste  system  implies 
an  hereditary  differentiation  of  people.  Such  systems, 
although  not  formally  prescribed,  are  quite  prevalent 
and  operate  in  the  choice  of  a  vocation  and  'similar 



situations  of  everyday  life.  The  interpretation  of  family 
resemblances,  and  even  in  some  cases  the  development 
of  family  groupings  themselves,  rest  upon  specific  under- 
lying hypotheses  regarding  the  causal  factors  in  human 
resemblance  and  dissimilarity. 


The  causes, of  individual  variation  are  to  be  sought  in 
the  individual's  hereditary  background  and  in  the  en- 
vironmental conditions  to  which  he  has  been  exposed. 
Every  trait  or  behavior  manifestation  of  the  individual 
depends  both  on  his  heredity  and  on  his  environment. 
Traits  and  activities  cannot  be  classified  into  those  which 
are  inherited  and  those  which  are  acquired.  The.pcp&lem 
thus  resolves  itself  into  a  determination  of  the  relative 
contribution  of  hereditary  and  environmental  factors  in 
the  development  of  the  individual.  To  what  extent  can 
the  development  of  any  given  characteristic  be  altered 
by  the  control  of  environmental  influences,  and  to  what 
extent  is  such  modification  limited  by  hereditary  condi- 
tions? Individual  variations  found  under  similar  heredi- 
tary conditions  may  be  attributed  to  the  operation  of 
different  environmental  factors.  Similarly,  when  the  en- 
vironments are  sufficiently  similar,  any  dissimilarity  of 
behavior  may  be  regarded  as  indicative  of  a  difference  in 
heredity.  We  may  speak,  therefore,  of  hereditary  and 
environmental  determiners  or  factors  in  the  development 
of  a  given  characteristic,  although  such  a  dichotomy 
cannot  be  applied  to  traits. 

How  does  heredity  operate  in  the  development  of  the 
individual?  The  understanding  of  the  mechanism  of 
heredity  has  been  greatly  furthered  by  the  concept  of  the 
gene.  The  individual  begins  life  at  conception  with  the 
union  of  one  germ  cell  from  each  parent,  the  ovum  of  the 


female  and  the  spermatozoon  of  the  male.  Each  of  these 
cells  contains  hundreds  of  thousands  of  very  minute  parti- 
cles, called  genes.  A  gene  is  the  carrier  of  a  "unit  char- 
acter," i.e.,  an  hereditary  factor  or  influence  which  always 
operates  as  a  unit,  or  in  an  all-or-none  fashion.  These 
unit  characters  of  the  geneticists  are  not  to  be  confused 
with  traits  as  ordinarily  conceived,  but  are  of  a  much 
more  elementary  nature.  Thus,  even  a  relatively  simple 
characteristic  such  as  eye  color  depends  upon  the  com- 
bined influence  of  a  very  large  number  of  separate  genes. 
Such  complex  hereditary  determination  would  of  course 
produce  varying  degrees  of  a  trait,  even  though  the 
individual  genes  may  be  characterized  only  by  presence 
or  absence. 

It  is  obvious  that  any  attempt  to  identify  psychological 
characteristics,  and  especially  such  a  manifold  and  ill- 
defined  phenomenon  as  "intelligence/5  with  unit  chaty 
acters  is  entirely  inconsistent  with  the  concepts  and  data 
of  genetics.  Nevertheless  such  an  attempt  was  actually- 
made  in  the  zeal  which  followed  the  early  spread  of  menta} 
tests.1  The  experimental  location  of  the  specific  genes 
which  contribute  towards  the  development  of  observable 
traits  of  the  organism  is,  on  the  other  hand,  an  extremely 
difficult  task.  In  recent  years  rapid  strides  have  been 
made  along  these  lines  by  geneticists,  as  is  illustrated  by 
the  excellent  work  of  Morgan  (18)  on  the  gene  "maps" 
of  the  fruit  fly,  Drosophila  melanogaster.  This  phase  of 
the  geneticist's  work  is,  however,  still  in  its  infancy. 
The  data  available  at  present  are  too  meagre  to  be  of 
any  di'rect  assistance  in  the  understanding  of  complex 
behavior  phenomena. 
f  The^ere^ 
found  in  the"  almost  unlimited  variety  of  possible  #ene 

,       ,  -  ',  ,  ,  ,      „  .       •     '  '  f    <*  ¥fr*   ,*  n    »..«.  ,-^*'<»  <f   "•<-«    *.«,  n>«  f^tmf 

'  ' l  Cf.  Goddard  (8),  especially  Chs,  VII  and  VIIL 


combinations  which  presents  itself,  especially  in  the  case 
of  such  a  complex  organism  as  man.  When  male  and 
female  germ  cells  unite  at  conception,  the  fertilized  ovum 
contains  two  sets  of  genes,  one  from  each  parent.  These 
genes  combine  in  a  variety  of  ways  in  each  of  the  daughter 
cells,  and  hence  the  different  germ  cells  of  the  resulting 
organism  will  not  be  identical  in  their  gene  constitution. 
The  individual  offspring  of  this  organism  will  therefore 
differ  among  themselves  because  they  have  developed 
from  different  germ  cells.  When  we  add  to  this  the  fact 
that  such  offspring  result  from  the  union  of  germ  cells 
from  two  parents,  each  of  whom  has ,  germ  cells  which 
differ  among  themselves,  the  number  of  possible  varia- 
tions becomes  very  large.  Thus-  it  is  small  wonder  that 
duplicate  individuals  are  not  produced  by  chance. 

TJie.  only  exception  to  this  individual  diversity  of  "gene 
constitution  is  that  of  identical  twins,  which  develop  from 
a^  single  fertilized  ovum.  Such  twins  are  always  of  the 
same  sex  and  identical  in  appearance.  Fraternal  twins, 
on  the  other  hand,  do  not  reveal  such  close  resemblance 
and  may  be  either  of  the  same  or  opposite  sex.  The  heredi- 
tary similarity  of  fraternal  twins  is  no  greater  than  that 
of  ordinary  siblings,1  since  they  result  from  the  simul- 
taneous development  of  two  different  cells.  Fraternal 
twins  are,  however,  exposed  to  the  same  prenatal  en- 
vironment, which  may  be  an  important  factor  in  render- 
ing them  similar  in  their  subsequent  development. 

Certain  popular  misconceptions  regarding  the  mani- 
festations of  heredity  should  be  cleared  up  at  the  outset. 
In  the  first  place,  there  is  a  ,c,pmmon  belief  thatjnj^^tjance 

...".....,.  v^,,^*-***"***"^  »**•**"»  <  *   ""    "*     "'    "'       ,.».'» ^f^f'.-M'ti"'     mmm  ^  _ 

is  indicated  only  by  resembla^ 

ancestors.    This  is  shown  to  be  false  by  a  consideration 

of  the  mechanism  of  inheritance.     The  germ  plasm  is 

1  Siblings  is  a  general  term  employed  to  cover  both  brothers  and  sisters. 


continuous  from  generation  to  generation  and  is  not  "pro- 
duced" by  the  individual  parents.  The  parents  simply 
transmit  this  germ  plasm  to  the  offspring.  Hence  the  indi- 
vidual inherits  from  all  of  his  direct  ancestors,  and  not 
only  from  his  parents.  Some  characteristic  which  may 
have  been  latent  for  many  generations  may  become  domi- 
nant because  of  a  particular  combination  of  genes  and 
the  result  will  be  an  individual  very  unlike  his  parents 
or  immediate  forbears  in  some  one  respect.  Instances  of 
this  sort  are  not  uncommon  in  family  histories.  In  such 
a  case,  heredity  actually  serves  to  make  the  child  unlike 
hisi,  parents. 

A  further  possible  source  of  misunderstanding  is  to  be 
found  in  the  common, habit  of  speaking  about  functions 
and  activities  as  determined  by  hereditary  factors.  ^Hered- 
ity can  only  exert  a  direct  control  over  the  development 
of  structures.  Insofar  as  a  given  activity  involves  the 
presence  of  certain  structures,  such  as  vocal  organs,  hands, 
glands,  nervous  system,  the  hereditary  factors  underlying 
the  development  of  these  structures  will  influence  activity. 
Likewise,  the  nature  and  degree  of  development  of  organs 
will  affect  their  functions.  But  this  is  only  a  limiting  con- 
dition imposed  upon  the  development  of  a  given  type  of 
behavior.  Htei^dijtl!;,r£jactors  may  prevent  the  appearance 
of  a  function  through  the  aBsence  of  the  necessary  struc- 
tures, but  the  converse  does  not  hold.  The  presence  of 
such  structures  does  not  imply  that  the  particular  form  of 
behavior  in  question  will  appear. 

Another  popular  notion  occasionally  met  is  that  what- 
ever is  present  at  birth  is  inherited  and  whatever  appears 
subsequently  is  acquired.  As  it  stands,  this  statement  is 
inconsistent  with  the  fundamental  concept  of  development 
as  an  interaction  of  hereditary  and  environmental  influ- 
ences, Even  if  reworded  to  eliminate  this  inconsistency, 


however,  it  is  still  misleading.  Thus  it  is  equally  false 
to  think  of  the  operation  of  hereditary  influences  in  the 
development  of  any  trait  as  ceasing  at  birth  or,  conversely, 
to  think  of  environmental  influences  as  beginning  to  oper- 
ate at  birth.  Hereditary  factors  may  influence  the  de- 
velopment of  the  individual  long  after  birth,  and  in  fact 
throughout  the  life  span.  Even  the  onset  of  death  itself 
may  be  determined  partly  by  hereditary  factors,  as  sug- 
gested by  the  observation  that  longevity  or  "long  life" 
tends  to  run  in  families.  Hereditary  influences  may  be- 
come manifest  for  the  first  time  at  any  age.  Similarly, 
environmental  influences  begin  to  operate  upon  the  indi- 
vidual from  the  moment  of  conception.  The  importance 
of  temperature,  chemical,  and  other  types  of  stimulation 
in  the  prenatal  surroundings  of  the  developing  embryo  is 
rapidly  coming  to  be  recognized.  Birth  is  not  to  be  re- 
garded as  either  a  beginning  or  an  end  in  the  life  of  the 
organism,  but  as  a  relatively  minor  occurrence  in  a 
developmental  continuum  which  begins  at  conception  and 
ends  at  death. 


In  general  it  may  be  said  that  the  earlier  an  environ- 
mental condition  operates  in  the  life  of  the  individual,  the 
more  pronounced  will  be  its  effect.  After  an  advanced 
state  of  growth  has  been  reached,  the  organism  becomes 
much  less  modifiable.  For  this  reason,  the  stimuli  to 
which  the  individual  is  exposed  during  the  embryonic 
stage  exert  a  pronounced  and  lasting  influence  upon  its 
future  development.  Variations  in  diet  and  nutrMon, 
glandular_secretion,  and  other  "conditions  of  the  mother 

blood  have  a  marked »,  ,dfe£t»  uppn^  ^fll.^?i^^^^l^' 
tHe  ejcabi-y^*  The  structural  development  of  the  individual 


is  definitely  influenced  by  such  environmental  factors. 
It  is  also  possible  that  a  certain  amount  of  rudimentary 
learning  may  occur  during  prenatal  life.  The  presence  of 
-certain  reflexes  and  other  simple  movements  in  young 
embryos  has  been  definitely  established.  Such  responses 
may  early  become  conditioned  in  various  ways  to  changes 
of  temperature,  pressure,  and  other  stimuli  furnished  by 
the  intra-uterine  environment.  The  study  of  prenatal 
behavior  opens  interesting  possibilities,  although  so  far 
it  offers  only  very  meagre  information.  In  the  field  of 
structural  development,  however,  the  data  are  much  more 

Many  instances  of  experimentally  induced  structural 
changes  in  lower  animals  have  been  reported  from  time 
to  time.  A  very  curious  transformation  can  be  produced 
in  the  axolotl,  a  large  salamander  (cf.  14,  pp.  117,  124- 
125).  Normally,  this  animal  has  prominent  external  gills, 
a  large  tail  adapted  for  swimming,  and  other  character- 
istics suited  for  life  in  the  water.  If  the  young  axolotl 
is  fed  on  thyroid,  it  loses  its  gills,  and  its  body  becomes 
generally  altered  so  that  it  is  no  longer  adapted  to  swim- 
ming. The  animal  then  becomes  a  land  salamander, 
known  as  Amblystoma,  and  returns  to  the  water  only 
to  lay  its  eggs. 

In  the  fruit  fly,  a  defective  gene  causes  the  animal  to 
produce  "reduplicated  legs,"  i.e.,  certain  joints  of  the 
legs,  or  entire  legs,  are  doubled.  Although  the  inheritance 
of  this  defective  gene  has  been  definitely  traced,  this 
characteristic  will  not  appear  under  certain  environmental 
conditions  (12).  If  animals  known  to  have  the  defective 
gene  are  kept  at  a  sufficiently  warm  temperature,  the 
additional  leg  or  joint  will  not  develop.  Successive  gen- 
erations bred  under  these  conditions  will  have  a  normal 
appearance.  If,  however, '  any  of  their  offspring  are  al- 


lowed  to  develop  in  colder  temperatures,  the  defect  will 
reappear.  This  furnishes^,defimte~  illustration  of ~±ke  iact 
that  even  a  clearly  demonstrable  " inherited  defect"  is 
actually  only  a  tendency  to  develop  in  a  given  way  under 
certain  environmental  conditions. 

Experimentally  produced  "  monsters  "  furnish  very  strik- 
ing examples  of  the  influence  of  prenatal  environment 
(24,  Chs.  VI  >and  VII).  In  experiments  on  fish  eggs, 
" Siamese  twin"  fish  have  been  produced  by  artificially 
inhibiting  or  slowing  down  the  rate  of  development  at  an 
early  age  through  cold,  insufficient  oxygen,  or  the  applica- 
tion of  ultra-violet  rays.  In  some  cases,  one  twin  is  much 
smaller  than  the  other  and  is  deformed,  the  larger  twin 
being  a  perfectly  normal  fish.  Two-headed  monsters 
have  been  produced  among  tadpoles  and  several  species 
of  fish  by  the  application  of  various  chemical  or  mechan- 
ical stimuli.  Striking  variations  in  the  number  and  posi- 
tion of  the  eyes  of  minnows  have  likewise  been  induced. 
If  the  eggs  of  the  minnow  are  allowed  to  develop  in  sea 
water  to  which  has  been  added  an  excess  of  magnesium 
chloride,  peculiar  eye  conditions  will  appear  in  a  large 
majority  of  the  embryos.  Instead  of  the  usual  two  eyes, 
many  will  develop  a  centrally  placed  "cyclopean"  eye, 
so  named  after  the  one-eyed  Cyclops  of  mythology.  Others 
may  show  a  single  lateral  eye,  placed  to  the  right  or  left 
of  the  head.  Or  the  two  eyes  may  be  abnormally  close 
together.  Other  physical  or  chemical  agents  may  be 
employed  to  produce  the  same  anomalies  of  development. 
The  determining  factor  in  the  development  of  a  particular 
abnormality  seems  to  be  the  stage  at  which  the  agent  is 
introduced,  rather  than  the  nature  of  the  specific  agent 
employed.  The  essential  effect  is  a  change  in  the  rate  of 
development,  which  alters  the  balance  of  growth  among 
the  different  parts  of  the  organism. 


Thus  we  cannot  even  speak  of  certain  structural  char- 
acteristics as  being  " normal"  for  a  given  species  and  fixed 
by  hereditary  constitution.  If  the  environment  in  which 
the  organisms  develop  were  to  undergo  a  change  of  a  more 
or  less  permanent  nature,  a  different  set  of  characteristics 
would  come  to  be  considered  normal.  Similarities  of 
development  are  attributable  to  common  exposure  to  an 
essentially  similar  environment  as  much  as  to  the  pos- 
session of  common  genes.1 


Numerous  experiments  have  shown  the  possibility  of 
pronounced  alteration  in  behavior  as  a  result  of  environ- 
mental differences.  Animals  reared  in  isolation  from 
other  members  of  the  species  or  from  individuals  of  the 
opposite  sex,  or  in  close  association  with  a  human  child, 
have  developed  curious  modifications  of  behavior.  Activ- 
ities which  are  commonly  assumed  to  be  "unlearned"  and 
fixed  by  hereditary  constitution  have  proved  susceptible 
to  marked  change.  Universal  characteristics  in  behayior, 
aajn  structure, ,..h,aye  Be^a  shown 'to  result  as  much  frqm 
common^environments  and  similar  opportunities  for  learn- 
ing^as  from^common  heredity.  A  few  illustrations  of 
experimental  attempts  to  alter  behavior  traits  will  be 
considered.  It  is  understood,  of  course,  that  by  alteration 
is  here  meant  simply  the  production  of  behavior  different 
from  that  which  develops  in  the  ordinary  environment. 
There  is  no  implication  that  one  kind  of  behavior  is 
fundamentally  any  more  "natural"  than  another. 

The  songs  of  different  birds,  generally  considered  so 
characteristic  of  the  species,  have  been  found  to  contain 

1 A  very  comprehensive  survey  of  data  showing  the  effect  of  environmental 
factors  upon  the  development  of  stature  and  weight  in  the  human  is  to  be  found 
in  Sanders  (21). 


much  that  is  learned  in  their  specific  manifestations.  A 
group  of  newly  hatched  orioles  were  segregated  from  all 
older  members  of  the  species  and  brought  up  by  them- 
selves (22).  When  the  birds  reached  a  certain  age,  they 
began  to  sing.  Their  song,  however,  was  not  the  character- 
istic oriole  song,  but  a  new  one.  When  other  newly 
hatched  orioles  were  reared  with  the  group  that  had  grown 
up  in  the  laboratory,  the  young  birds  in  turn  learned  the 
new  song  developed  by  the  first  group. 

Similarly,  a  sparrow  reared  in  a  nest  of  canaries  gradu- 
ally abandoned  his  own  chirps  and  began  to  take  over 
the  canary's  call  note  (i).  After  about  a  month  of  as- 
sociation with  the  canaries,  the  sparrow  began  to  imitate 
their  song.  Although  at  first  very  harsh  and  confused, 
this  song  eventually  developed  into  a  genuine  canary-like 
performance.  These  and  similar  experiments  have  shown 
that  although  birds  will  develop  some  sort  of  song  when 
they  reach  a  certain  stage  of  structural  development, 
the  specific  nature  of  the  song  depends  upon  environ- 
mental factors. 

Sexual  behavior  has  also  proved  to  be  dege^deijtjjgpn 
""^  manifestations.  Some  4  form  of 

occur  at  a  definite  developmental 

stage,  because  of  the  presence  of  glandular  secretions  in 
the  blood  and  other  physiological  factors.  Thg  particular 
way  in  which  such  activity  is  expressed  and  the  object 
towards  which  it  is  directed,  however,  will  vary  according 
to  environmental  circumstances.  In  an  experiment"  on 
male  doves  reared  in  isolation  from  other  members  of 
the  species,  a  number  of  sexual  "abnormalities"  were 
observed  (3).  The  birds  would  bow  and  coo  to  the  experi- 
menter as  normal  birds  do  to  members  of  their  own 
species.  They  seemed  to  pay  especial  attention  to  the 
experimenter's  hand  with  which  they  came  into  contact 


when  fed;  one  bird  actually  went  through  the  act  of 
copulation  while  on  the  hand  taking  food.  Female  doves 
reared  in  isolation  developed  similar  anomalies  of  be- 
havior (2).  If  the  experimenter  stroked  them  and  preened 
the  feathers  of  their  head  and  neck,  they  exhibited  char- 
acteristic courting  behavior.  Egg-laying  was  actually 
induced  in  many  instances  by  this  method.  Experimental 
"homosexuality"  was  produced  in  a  large  number  of 
cases  when  two  female  pigeons  were  reared  together. 
In  such  cases,  the  animals  would  display  the  usual  court- 
ing performance  towards  each  other,  egg-laying  then 
following  in  both. 

Equally  curious  variations  of  behavior  were  noted  in  a 
young  monkey  separated  from  its  mother  at  the  age  of 
three  days  and  brought  up  in  isolation  from  all  members 
of  the  species  during  the  first  18  months  of  life  (5,  6). 
The  development  of  sexual  behavior  in  general  was  mark- 
,edly  delayed.  During  the  period  of  isolation  there  was  a 
minimum  of  the  sex  behavior  ordinarily  displayed  by 
monkeys  at  that  age.  At  the  age  of  18  months,  the  period 
of  isolation  was  discontinued,  and  the  monkey  was  sub- 
sequently brought  up  with  other  members  of  the  species. 
At  this  time,  sex  behavior  began  to  appear,  but  in  a  very 
rudimentary  form.  Attempts  at  copulation  were  very 
crude  and  trial-and-error  was  exhibited.  Sexual  activity 
was  shown  indiscriminately  towards  males  and  females, 
•as  well  as  towards  monkeys  of  other  species,  rags  and 
other  soft  objects,  and  the  experimenter's  arm  and  hand. 
With  continued  association  with  other  members  of  the 
species,  -normal  sexual  activity  eventually  developed. 
Other  forms  of  behavior,  such  as  feeding,  play-activity, 
and  grooming,  were  also  affected  by  the  prolonged  period 
of  isolation. 

The  recent  investigation  of  Kellogg  and  Kellogg  (16), 


in  which  a  young  chimpanzee  was  reared  for  a  short 
period  in  a  typically  human  environment,  throws  further 
light  upon  the  factors  affecting  behavioral  development. 
The  chimpanzee,  a  female  named  "Gua,"  was  isolated 
from  its  mother  at  the  age  of  7!  months  and  brought  up 
in  the  company  of  the  investigator's  own  son,  Donald, 
then  10  months  old.  The  association  was  continued  for 
a  period  of  nine  months.  The  chimpanzee  was  not  treated 
as  a  pet,  but  as  a  child,  and  the  two  subjects  were  given 
as  nearly  as  possible  identical  care.  Gua  was  clothed  in 
the  same  manner  as  the  child,  and  showed  no  difficulty 
in  keeping  on  shoes,  stockings,  and  other  common  articles 
of  clothing.  She  slept  in  a  bed  with  the  usual  accessories, 
such  as  sheets  and  blankets.  Excellent  progress  was 
made  by  the  chimpanzee  in  learning  to  eat  with  a  spoon 
and  drink  out  of  a  glass.  She  was  able  to  manipulate 
pencil  and  paper  and  produce  simple  scribblings.  Gua 
also  learned  to  respond  to  oral  language,  and  by  the 
termination  of  the  experimental  period  understood  over 
50  words  or  simple  phrases,  such  as:  "Blow  the  horn" 
(in  the  car);  "Show  me  your  nose";  "Do  you  want  to  go 
bye-bye?"  "Take  it  out  of  your  mouth."  The  degree  to 
which  it  proved  possible  to  "humanize"  the  behavior  of 
this  ape  is  indeed  suggestive,  especially  in  view  of  the 
fact  that  the  period  of  residence  in  the  human  environment 
was  of  relatively  short  duration  and  did  not  begin  at 

What  would  be  the  effect  of  bringing  up  a  human  child 
in  isolation  or  in  exclusive  contact  with  lower  animals? 
Several  cases  of  children  found  in  such  situations  are  on 
record.  The  child's  condition  and  behavior  repertory  at 
the  time  of  discovery,  as  well  as  its  subsequent  develop- 


ment  when  transferred  to  a  normal  environment,  furnish 
interesting  material  on  the  heredity-environment  question. 
The  most  famous  case  is  probably  that  of  the  Wild  Boy 
of  Aveyron,  as  it  has  come  to  be  known  in  the  psycho- 
logical literature.  In  September,  1799,  three  sportsmen 
came  upon  a  boy  of  II  or  12  in  a  French  forest.  The  boy 
was  completely  naked,  unkempt,  scarred,  unable  to  talk, 
and  seemed  to  have  been  leading  a  wild,  animal-like 
existence.  He  was  seized  by  the  men  as  he  was  climbing 
a  tree  to  escape  their  pursuit,  and  was  subsequently 
brought  to  civilization.  He  finally  came  under  the  guid- 
ance and  observation  of  the  French  physician,  Itard. 
The  very  illuminating  account  which  Itard  published  on 
his  own  findings  has  immortalized  the  Wild  Boy  of  Avey- 
ron.  When  found,  the  boy  seems  to  have  been  deficient 
in  all  forms  of  behavioral  development,  including  sensory, 
motor,  intellectual,  and  emotional.  This  is  clearly  brought 
out  in  the  following  brief  description  (13,  pp.  5-8), 

His  eyes  were  unsteady,  expressionless,  wandering  vaguely 
from  one  object  to  another  without  resting  on  anybody;  they 
were  so  little  experienced  in  other  ways  and  so  little  trained 
by  the  sense  of  touch,  that  they  never  distinguished  an  object 
in  relief  from  one  in  a  picture.  His  organ  of  hearing  was 
equally  insensible  to  the  loudest  noises  and  to  the  most 
touching  music.  His  voice  was  reduced  to  a  state  of  complete 
muteness  and  only  a  uniform  guttural  sound  escaped  him. 
His  sense  of  smell  was  so  uncultivated  that  he  was  equally 
indifferent  to  the  odor  of  perfumes  and  to  the  fetid  exhala- 
tion of  the  dirt  with  which  his  bed  was  filled.  Finally,  the 
organ  of  touch  was  restricted  to  the  mechanical  function  of 
grasping  objects.  Proceeding  then  to  the  state  of  the  intel- 
lectual functions  of  this  child,  the  author  of  the  report  pre- 
sented him  to  us  as  being  quite  incapable  of  attention  (except 
for  the  objects  of  his  needs)  and  consequently  of  all  those  op- 


erations  of  the  mind  which  attention  involves.  He  was  des- 
titute of  memory,  of  judgment,  of  aptitude  for  imitation,  and 
was  so  limited  in  his  ideas,  even  those  relative  to  his  immediate 
needs,  that  he  had  never  yet  succeeded  in  opening  a  door  or 
climbing  upon  a  chair  to  get  the  food  that  had  been  raised 
out  of  reach  of  his  hand.  In  short,  he  was  destitute  of  all 
means  of  communication  and  attached  neither  expression 
nor  intention  to  his  gestures  or  to  the  movements  of  his 
body.  He  passed  rapidly  and  without  any  apparent  motive 
from  apathetic  melancholy  to  the  most  immoderate  peals  of 
laughter.  .  ,  .  His  locomotion  was  extraordinary,  literally 
heavy  after  he  wore  shoes,  but  always  remarkable  because 
of  his  difficulty  in  adjusting  himself  to  our  sober  and  meas- 
ured gait,  and  because  of  his  constant  tendency  to  trot  and 
to  gallop.  He  had  an  obstinate  habit  of  smelling  at  anything 
that  was  given  to  him,  even  the  things  which  we  consider 
void  of  smell;  his  mastication  was  equally  astonishing,  ex- 
ecuted as  it  was  solely  by  the  sudden  action  of  the  incisors, 
which  because  of  its  similarity  to  that  of  certain  rodents, 
was  a  sufficient  indication  that  our  savage,  like  these  animals, 
most  commonly  lived  on  vegetable  products. 

It  is  interesting  to  note  that  the  sensory  deficiency  of 
this  boy  seems  to  have  been  quite  specific  and  in  many 
instances  directly  traceable  to  his  mode  of  life.  Thus 
Itard  observed  that  "the  sound  of  a  cracking  walnut  or 
other  favorite  eatable  never  failed  to  make  him  turn 
around  .  .  .  nevertheless,  this  same  organ  showed  itself 
insensible  to  the  loudest  noises  and  the  explosion  of  fire- 
arms" (13,  p.  15).  Sexual  development  showed  the  same 
general  undifferentiated  type  of  response  observed  in  the 
case  of  animals  reared  in  isolation.  Following  the  onset 
of  puberty,  periods  of  vague  restlessness  and  discomfort 
as  well  as  occasional  fits  of  sadness  or  anger  were  noted, 
without,  however,  the  development  of  specific,  normal 
sexual  activity. 


After  five  years  of  ingenious,  painstaking,  and  method- 
ical training,  Itard  abandoned  the  task,  having  failed 
to  bring  the  boy  up  to  normal.  This  has  led  many  to 
conclude  that  the  Wild  Boy  of  Aveyron  must  have  been 
an  imbecile  from  birth,  who  had  been  abandoned  by  his 
parents  because  of  his  mental  deficiency.  Such  a  conclu- 
sion, however,  overlooks  several  important  points.  In 
the  first  place,  marked  improvements  were  effected  by 
the  training,  even  though  a  normal  level  was  not  reached. 
For  example,  although  the  boy  could  not  learn  to  articu- 
late sounds,  he  succeeded  in  learning  simple  written  lan- 
guage, so  that  he  was  finally  able  to  reproduce  written 
words  from  memory  and  to  use  them  "to  express  his 
wants,  to  solicit  the  means  to  satisfy  them  and  to  grasp 
by  the  same  method  of  expression  the  needs  or  the  will 
of  others3'  (13,  p.  84).  Secondly,  had  the  boy  actually 
been  feebleminded,  he  should  very  probably  have  been 
unable  to  survive  in  the  very  trying  circumstances  of  his 
primitive  environment.  Finally,  the  fact  that  the  train- 
ing was  begun  so  late  in  life  may  furnish  an  adequate 
explanation  of  its  lack  of  success.  The  environment  of 
early  childhood  is  too  important  in  determining  subse- 
quent development. 

The  more  recently  discovered  "wolf  children"  of  India 
(23,  15)  represent  a  similar  case.  In  1921,  two  girls,  aged 
about  two  to  four  and  eight  to  nine,  respectively,  were 
found  living  in  a  cave  with  wolves  in  a  sparsely  settled 
region  of  India.  They  were  taken  into  a  local  orphanage 
where  some  attempts  were  made  to  train  them.  It  proved 
very  difficult,  however,  to  keep  the  girls  in  good  health, 
particularly  because  they  could  not  be  induced  to  eat  a 
normal,  varied  human  diet,  but  retained  the  feeding 
habits  they  had  acquired  from  association  with  carnivo- 
rous animals.  The  younger  girl,  "Amala,"  died  in  less 


than  a  month;  the  elder,  "Kamala,"  lived  for  about 
eight  years.  As  a  result,  most  of  the  observations  which 
have  been  recorded  were  made  on  the  older  child. 

Kamala,  like  her  younger  companion,  showed  a  great 
preference  for  raw  meat.  Although  never  known  to  kill 
any  domestic  animal,  she  was  fond  of  pouncing  upon 
any  killed  animal  which  she  found.  The  odor  of  meat 
could  be  detected  at  a  great  distance  and  a  keen  animal- 
like  sense  of  smell  was  generally  exhibited.  Hearing  was 
also  very  acute.  The  eyes  are  described  as  possessing  a 
peculiar  glare,  like  the  eyes  of  dogs  or  cats  in  the  dark. 
It  seemed  that  Kamala  could  see  much  better  at  night 
than  in  the  daytime,  and  she  seldom  slept  after  midnight. 
When  dressed,  she  tore  off  her  clothes.  Eating  and  drink- 
ing were  accomplished  by  lowering  the  mouth  into  the 
plate,  like  a  dog.  Eventually,  however,  she  was  taught 
to  use  her  hands  in  eating.  Locomotion  originally  con- 
sisted of  crawling  on  all  fours.  She  finally  learned  to 
walk  erect  on  two  legs,  although  she  was  never  able  to 
run  in  this  position.  She  was  accustomed  to  emit  a  cry 
or  howl  in  a  peculiar  voice,  neither  animal  nor  human. 
With  prolonged  training,  she  was  able  to  say  about  forty 
words  and  form  simple  sentences  of  two  or  three  words, 
although  the  original  howl  was  still  repeated  occasion- 

Mention  should  also  be  made  of  the  celebrated  and 
mysterious  case  of  Kaspar  Hauser  (cf.  26,  pp.  290-292), 
about  whom  so  much  has  been  written.  Some  accounts 
suggest  that  this  boy  was  an  heir  to  some  princely  house 
and  was  put  out  of  the  way  by  political  enemies.  He 
was  apparently  confined  from  early  childhood  in  a  dark 
cell,  not  large  enough  for  him  to  stand  upright.  No 
clothing  or  cover  was  furnished  except  a  shirt  and  trousers. 
When  he  awoke,  he  was  accustomed  to  find  bread  and 


water,  but  he  never  saw  the  person  who  brought  them 
and  he  had  no  knowledge  of  the  existence  of  other  living 
creatures  besides  himself.  He  was  released  in  1828,  when 
about  17  years  of  age.  At  this  time  he  was  first  discov- 
ered, wandering  aimlessly  about  the  streets  of  Nuremberg. 
He  could  not  talk,  but  repeatedly  uttered  certain  phrases 
meaninglessly.  He  is  reputed  to  have  had  a  remarkable 
sense  of  smell  and  a  surprising  ability  to  see  in  the  dark. 
His  walking  resembled  the  first  efforts  of  a  child.  After 
various  vicissitudes,  his  instruction  was  undertaken  by 
Prof.  Daumer.  Under  the  latter' s  tutelage,  Kaspar  Hauser 
made  rapid  progress,  and  soon  learned  to  speak.  By  this 
means  he  was  able  to  communicate  what  he  recalled 
of  his  life  in  the  cell  As  in  other  cases  of  children  brought 
into  contact  with  civilization  relatively  late  in  life.,  his 
education  never  brought  him  quite  up  to  normal. 

These  examples  illustrate  the  close  dependence  of  hu- 
man^deyelopme'nFupoii  the  environment  in  which  the 
subject  is  reared  and  the  type  of  stimulation  to  which 
he  Jsjsxpftsjpd.  If  a  child  is  deprived  of  normal  human 
contacts,  his  behavior"  will  come  to  resemble  in  many 
ways  that  of  a  low-grade  idiot.  Such  a  condition  has,  in 
fact,  been  regarded  as  a  sort  of  environmental  feeble- 
mindedness and  has  been  given  the  name  of  isolation 
amentia  (cf.  26,  pp.  285,  290).  When  a  child  is  brought 
"up  in  contact  with  animals,  striking  similarity  to  the 
behavior  of  those  animals  is  exhibited,  and  such  behavior 
proves  difficult  to  eradicate  once  it  has  become  firmly 
established.  Subsequent  educational  efforts  are  inadequate 
to  undo  the  effects  of  early  nurture.  Rousseau's  dream 
of  the  "  noble  savage "  whose  inner  nature  is  allowed  to 
develop,  free  and  unhampered  by  human  interference, 
proves  to  be  a  vain  chimera.  The  situation  has  been  very 
aptly  summarized  by  Stratton  (25,  p.  597): 


Lack  of  association  with  adults  during  a  certain  critical 
period  of  early  childhood,  it  seems  likely,  produces  in  some 
or  all  normal  children  marks  like  those  of  congenital  defect. 
The  evidence  seems  against  the  romantic  view  that  a  civ- 
ilized community  is  a  chief  obstacle  to  the  development  of 
personality.  On  the  contrary,  the  higher  forms  of  personality 
become  possible  only  in  and  through  such  a  community. 
By  our  biological  endowment  alone,  or  by  this  as  developed 
by  maturing  and  learning  in  an  infrahuman  environment, 
we  remain  man-beasts.  We  become  human  only  by  active 
intercourse  in  a  society  of  those  who  already  have  become 


The  environmental  variations  described  in  the  preced- 
ing sections  were  all  of  a  very  drastic  nature.  It  might 
be  objected  that  such  cases  tell  us  little  about  the  ordi- 
nary range  of  individual  differences  observed  in  every- 
day life.  Then,  too,  the  examples  cited  above  represent 
single  cases;  little  is  known  about  the  exact  antecedents 
of  the  children;  and  the  observations  were  not  in  most 
cases  very  well  controlled.  It  would  seem  difficult  to 
draw  conclusions  from  such  information  alone.  The  facts 
of  human  development  suggested  by  these  strange  cases 
have,  however,  been  corroborated  by  data  gathered  from 
various  other  sources.  The  comparison  of  children  brought 
up  in  different  social  and  cultural  classes  or  occupational 
levels  presents  in  a  much  milder  form  the  same  type  of 
environmental  dissimilarity  and  concomitant  behavior 
variations  found  in  the  cases  of  "wild  children."  Such 
investigations  have  ordinarily  dealt  with  very  large  groups 
and  have  arrived  at  their  results  by  the  use  of  standardized 

Among  the  most  clear-cut  findings  on  the  effect  of  home 
environment  and  schooling  upon  intelligence  as  meas- 


ured  by  current  tests  are  those  of  Gordon  (10)  on  canal- 
boat  and  gypsy  children  in  England.  Gordon's  report, 
niaHe  in  the  course  of  his  official  duties  as  Inspector  of 
Schools,  is  based  on  the  Stanford-Binet  I.Q.'s  and  educa- 
tional test  scores  of  various  groups  of  children  whose 
schooling  is  deficient.  The  canal-boat  children  were  ob- 
tained from  two  special  schools  maintained  for  such 
groups.  Their  only  opportunity  to  attend  school  occurs 
when  the  canal-boats  are  tied  up  for  loading  or  discharg- 
ing. It  has  been  estimated  that  the  average  school 
attendance  of  canal-boat  children  is  only  5%  of  that  in 
ordinary  elementary  schools.  The  majority  are  only 
able  to  attend  school  about  once  a  month  for  one  or  two 
consecutive  half-days.  Their  home  surroundings,  although 
satisfactory  in  respect  to  conditions  of  health  and  cleanli- 
ness, are  intellectually  of  a  very  low  order.  Many  of 
the  adults  are  themselves  illiterate,  and  each  family  leads 
a  relatively  isolated  existence,  with  a  minimum  of  social 

The  average  LQ.  of  the  entire  group  of  76  canal-boat 
children  was  69.6.  Taken  at  face  value,  this  would  sug- 
gest at  best  a  borderline  group,  with  a  few  distinctly 
feebleminded  individuals.  Gordon's  analysis  of  the  data, 
however,  brings  out  very  vividly  the  influence  of  .the 
restricted  social  and  educational  facilities.  Thus  the  cor- 
reratfon^betweea  agp,  and  LQ.  within  the  entire  group 
proved  to  be  -.755.  This  indicates  a  marked  and  con- 
sistent tendency,  for,  the  older  children  to  get  lower  I.Q.'s 
than  t&e  younger.  Such  a  finding  is  quite  contrary  to  the 
data  onJ  growth  of  intelligence  in  the  usual  child.  The 
discrepancy  can  only  be  understood  in  terms  of  specific 
environmental  influences.  The  intellectual  environment 
of  the  younger  canal-boat  children  is  not  as  far  below 
normal  as  that  of  their  older  brothers  and  sisters.  The 


younger  child  in  any  home  is  exposed  to  relatively  simple 
intellectual  stimulation;  as  the  child  grows  older,  differ- 
ences in  schooling  and  in  the  cultural  level  of  the  home 
become  much  more  apparent. 

The  high  negative  correlation  with  age  is  corroborated 
by  analysis  of  individual  scores.  In  22  cases,  two  or  more 
children  from  the  same  family  were  tested.  With  only 
one  or  two  exceptions,  there  was  found  a  consistent  drop 
inT^O-  from  the  youngest  to  the  eldest  child  within  each 
family.  Most  of  the  youngest  children  had  I.Q.'s  between 
90  and  100,  which  would  place  them  within  the  normal 
group;  among  the  eldest,  on  the  other  hand,  were  several 
whose  I.Q.'s  were  low  enough  to  make  them  appear  dis- 
tinctly feebleminded.  A  further  corroborative  fact  brought 
out  by  this  analysis  is  that  the  mental  ages  of  children 
within  a  single  family  tended  to  be  very  similar,  even 
though  their  chronological  ages  differed.  Such  a  mental 
age  might  well  represent  the  limit  of  intellectual  develop- 
ment which  was  made  possible  by  the  available  educational 
opportunities  and  the  type  of  home  environment  furnished 
within  the  given  family. 

Gordon's  report  on  gypsy  children  lends  further  support 
to  the  interpretations  offered  above.  A  t:Qtil  of  M82  gypsy 
children  attending  four  schools  werq  .given  .the  Staafprd- 
BinetV  The^  school  attendance  of  gypsy  children,  although 
still  below  normal,  is  better  than  that  of  canal-boat  chil- 
dren. Actual  records  on  the  group  investigated  showed 
that  the  average  school  attendance  from  the  age  of  five 
to  the  time  of  testing  was  only  34.9%  of  the  total  number 
of  possible  school  days.  Living  conditions  were  in  general 
quite  crude  and  primitive.  The  groups  led  a  nomadic 
existence,  having  a  fixed  home  only  during  a  few  winter 
months;  it  was  at  this  time  that  the  children  attended 
school,  although  even  then  attendance  was  irregular.  The 


gypsy  children,  however,  mixed  more  frequently  with  other 
people  and  had  more  social  contacts  than  the  canal-boat 

The  average  I.Q.  of  the  entire  group  of  82  gypsy  children 
was  74.5.  A  significant  positive  correlation  of  .368  was 
found  between  I.Q.  and  percentage  of  school  attendance. 
Those  children  whose  school  attendance  had  been  less 
regular,  then,  tended  on  the  whole  to  have  lower  I.Q.'s. 
As  in  the  case  of  the  canal-boat  children,  a  significant 
negative  correlation  was  found  between  I.Q.  and  age, 
in  this  group  the  correlation  being  —.430.  Analysis  of 
siblings  within  each  family  also  showed  a  consistent  de- 
crease in  I.Q.  with  increase  in  age.  It  is  significant  that 
the  chief  exceptions  to  this  trend  occurred  in  those  families 
in  which  there  had  been  a  high  percentage  of  school 

The  findings  of  Gordon  on  both  groups  seem  to  point 
quite  irrevocably  to  factors  of  home  environment  and 
schooling  in  the  development  of  intelligence  as  commonly 
measured.  The  evidence  indicates  in  many  ways  that 
the  condition  of  mental  deficiency  was  not  present  at  the 
outset..  These  were  not,  as  might  be  supposed,  select 
groups,  in  the  sense  that  the  duller  persons  are  by  a 
gradual  process  attracted  into  the  life  of  the  gypsy  cara- 
van or  the  canal-boat  and  their  offspring  will  in  turn 
tend  to  be  duller  than  average.  Such  an  interpretation 
is  inconsistent  with  many  of  the  facts  given  above.  First, 
the  younger  children  were  on  the  whole  of  normal  intelli- 
gence; secondly,  the  older  the  child  and  the  longer  the 
differential  effects  of  inferior  home  and  school  conditions 
had  operated,  the  lower  the  I.Q.;  thirdly,  a  significant 
positive  correlation  was  found  between  regularity  of  school 
attendance  and  I.Q.;  fourthly,  the  gypsy  children  as  a 
group  had  higher  I.Q.'s  and  at  the  same  time  a  much 


higher  percentage  of  school  attendance  than  the  canal- 
boat  children.  Nor  can  the  greater  school  attendance 
of  the  former  be  attributed  to  more  intelligent  and  pro- 
gressive parents,  since  all  too  frequently  the  local  author- 
ities found  it  necessary  to  force  the  children  to  attend 
school  during  their  brief  winter  periods  of  stable  residence. 
Numerous  studies  have  appeared  on  the  relationship  of 
occupational  level  to  intelligence.  Large  groups  of  children 
have  been  tested  and  classified  according  to  father's  oc- 
cupation. The  latter  has  been  taken  as  an  objective, 
easily  obtained,  and  fairly  reliable  index  of  general  cul- 
tural level.  All  of  these  studies  have  shown  consistent 
differences  in  the  average  intelligence  of  children  in  vari- 
ous broad  occupational  groupings.  Such  data  are  not, 
however,  unambiguous.  It  is  difficult  to  determine,  with- 
out probing  further  into  the  particular  circumstances  in 
each  case,  which  is  cause  and  which  is  effect.  Several 
investigators  have  argued,  for  instance,  that  the  Tntel- 
lectual  differences  found  today  among  occupational  groups 
testify  to  a  gradual  hereditary  differentiation  which  ,has 
been  going  on  through  selection.  Thus  the  t  higher,  more 
intellectual  positions  come  to  be  filled  by  the  more  in- 
telligent individuals.  Since  intellectually  superior  parents 
tend  to  have  intellectually  superior  offspring,  the  children 
in  the  higher  occupational  strata  will  be  more  intelligent, 
on  the  whole,  than  those  from  the  semi-skilled  or  unskilled 
labor  classes.  The  alternative  hypothesis  explains  the 
intellectual  development  of  the  child  in  terms  of  the 
cultural  level  in  which  he  is  brought  up.  Thus  the  child 
who  grows  up  in  the  home  of  a  construction  laborer  does 
not  have  the  opportunities  for  intellectual  development 
and  consequently  will  not  attain  the  same  mental  level 
as  a  child  of  equal  initial  capacity  brought  up  in  the  home 
of  an  eminent  scientist  01  author.  With  these  alternative 


interpretations   in  view,   some  of  the  relevant  evidence 
will  be  examined. 

In  Great  Britain,  two  extensive  surveys  were  conducted 
with  an  intelligence  scale  of  British  construction.  In 
one  study  (4),  data  were  collected  on  13,419  children 
attending  elementary  and  secondary  schools  in  Nor- 
thumberland. The  other  study  (17)  was  conducted  on 
2047  elementary  schoolchildren  in  the  Isle  of  Wight.  In 
both  studies,  only  children  between  the  ages  of  n  and  13 
were  included.  The  children  were  classified  into  a  large 
number  of  groups  on  the  basis  of  specific  paternal  occupa- 
tion. More  general  trends  are  revealed,  however,  when 
the  occupations  are  grouped  into  certain  major  categories, 
The  average  I.Q.  as  well  as  the  number  of  children  in 
each  category  have  been  reproduced  below.  The  data 
of  the  two  studies  are  here  reported  together  (4,  p.  195; 
17,  p.  127). 

Northumberland  Isle  of  Wight 

Study  Study 


N        Average  N        Average 

IN  I.Q.  I.Q. 

A.  Professional  137  112.2  8  106.6 

B.  Managers  92  no.o  21  108.7 

C.  Higher  commercial  class  368  109.3  83  103.3 

D.  Army,  navy,  police,  postmen  129  105.5  141         99.9 

E.  Shopkeeping  class  748  105,0  224  100.7 

F.  Engineers  571  102.9  in  100.8 

G.  Foremen  256  102.7  17  103.3 
H,  Building  trades  717  102.0  207         99.1 
I.  Metal  workers,  ship-builders  820  100.9  I23         99-3 
J.  Miscellaneous  industrial  workers  472  100.6  201          99.1 
K.  Mining  and  quarrymen  5968  97.6  6         97.9 
L.  Agriculture,  all  classes  1128  97.6  277         96.7 
M.  Low-grade  occupations,  laborers  1214  96.0  328         96.0 
N.  Seamen,  bargemen,  boatmen,  etc.  60         97.4 

The  differences,  though  small,  show  a  certain  degree  of 
consistency.    This  is  particularly  apparent  if  the  upper 


seven  categories  In  the  list  are  compared  with  the  lower 
eight.  The  latter  are  predominantly  manual  occupations, 
whereas  the  former  depend  primarily  upon  intellectual, 
clerical,  or  executive  functions. 

In  America,  similar  investigations  have  been  conducted 
on  children  of  all  ages.  Goodenough  (9)  administered 
the  Kuhlman-Binet  tests  to  380  children  between  the 
ages  of  18  and  54  months.  Paternal  occupations  were 
classified  into  six  commonly  employed  categories.  The 
average  I.Q.'s  of  children  within  each  group  are  presented 

Occupational  Group  Average  I.Q.1 

I.  Professional  125.0 
II.  Semi-professional  (accountants,  drafts- 
men, etc.,  and  managerial)  *i9-7 

III.  Clerical  and  skilled  trades  113.4 

IV.  Semi-skilled  trades  and  minor  clerical  108.0 
V.  Slightly  skilled  107.4 

VI.   Unskilled  labor  95.8 

Similar  data  on  elementary  and  high  school  students 
are  to  be  found  in  a  study  by  Higgejtj^and  Nash  (n). 
The  Haggerty  Delta  2,,  a  common,  group  intelligence  test, 
was  administered  to  children  in  grades  3  to  8  and  high 
school  in  rural  districts  and  small  towns  of  New  York 
State.  The  authors  demonstrate  that  the  sampling  thus 
obtained  is  quite  representative  of  the  country  at  large 
in  respect  to  occupations.  The  number  of  cases  and 
median  and  range  of  I.Q.  found  within  each  occupational 
group  are  given  below  (n,  pp.  569-570).  The  data  for  ele- 
mentary and  high  school  groups  have  been  kept  sepa- 

1  The  LQ.'s  obtained  by  the  children  on  the  second  administration  of  the  test 
are  given  here,  as  these  were  shown  by  Goodenough  to  be  more  significant  than 
the  I.Q.'s  on  the  first  trial. 



•n  6688  Elementary 

Data  on  143$  High  School 


School  Children 




Range  in 


Range  in 










2O  I 




Business  and  cler- 






































1  06 


In  two  studies,  covering  both  high  school  and  all  grades 
of  the  elementary  school,  Pressey  (19,  20)  attempted  to 
show  the  dependence  of  occupational  differences  in  in- 
telligence upon  hereditary  factors.  The  children  were  all 
taken  from  a  small  middle  western  city  with  a  population 
of  about  12,000  and  no  foreign  element.  In  the  earlier 
study,  all  children  between  the  ages  of  10  and  14  inclusive 
were  given  the  Pressey  Group  Intelligence  Test;  the  total 
number  of  cases  obtained  was  548  and  ranged  academically 
from  the  third  grade  through  the  high  school.  The  later 
study  was  conducted  on  337  children  between  the  ages 
of  6  and  8  and  in  grades  one  to  three.  The  latter  group 
was  examined  with  the  Pressey  Primer  Scale,  a  non- 
verbal test.  Pressey  argued  that  since,  in  the  first  place, 
the  second  group  consisted  of  much  younger  children 
upon  whom  differential  home  environment  had  had  less 
chance  to  operate,  and  since  secondly,  the  test  employed 
was  non-verbal,  the  differences  in  intelligence  test  per- 
formance between  the  occupational  groups  could  be  at- 
tributed chiefly  to  heredity.  If  environment  had  been 
theumajor  cause,  the  occupational  ..differences, in.  .intelli- 
gence found  in  the  first  study  on  older,  children -should 
either  disappear  entirely. or  be  greatly-miHimized  in  the 
secbmf  study.  The  percentage  of  children  in  each  group 


whose  scores  fell  above  the  median  for  their  age  is  reported 
below  (20,  p.  369;  19,  p.  94). 

Percentage  above  Age  Median  (for  the 

~  ,  n  Total  Group) 

Occupational  Group  ^ 

Older  Group  Younger  Group 

(Ages  10-14)  (Ages  6-8) 

1.  Professional  85  "                            79 

2.  Executive  68  *6o 

3.  Artisan  41  54 

4.  Laborer  39  38 

The  findings  of  these  studies  are  typical  of  investiga- 
tions on  occupational  intelligence.  The  interpretations 
offered  by  the  authors  have  frequently  stressed  heredity 
as  the  underlying  cause  of  the  existing  differences.  What 
do  the  data  themselves  suggest?  A  number  of  facts 
should  be  noted  before  this  question  is  answered.  First, 
the_u.sL<SL.of. nonverbal -tests  such  as, .the  Kuhlman-jBIHet 
and  the  Pressey  Primer  Scale  does  aot  in  ..itself  rule  out 
environmental  influences.  Verbal  tests,  to  be  sure,  are 
more  directly  affected  by  specific  training  than  are  non- 
verbal, but  the  latter  still  imply  certain  stimulating  cir- 
cumstances of  which  the  child  in  the  poorer  home  may 
be  deprived.  Confidence  in  the  use  of  paper  and  pencil, 
adaptation  to  adults  outside  the  family,  practice  in  fol- 
lowing directions,  familiarity  with  pictures,  and  many 
similar  conditions  will  influence  the  child's  performance. 

Secondly,  it  is  impossible  tomake  a,umver,sal,«stateaient 
that  differences,,  resulting,  from  environmental  lackers  4n~ 
crease  with  age,  whereas  those  owing  principally  j:,Q.,.h£red- 
itafy  factors  are  manifested  from  a-ver^eaziy-age.  This 
was  found  to  be  the  case  in  the  groups  of  canal-boat  and 
gipsy  children  investigated  by  Gordon.  It  is  not  true, 
however,  of  children  whose  school  attendance  is  normal, 
but  who  may  come  from  inferior  homes.  In  such  a  group, 


we  should  in  fact  expect  intellectual  inferiority  to  be 
present  at  the  preschool  level,  when  the  child  has  been 
exposed  only  to  the  inferior  home  surroundings;  as  the 
children  grow  older  and  the  influence  of  equal  schooling 
becomes  effective,  the  original  differences  may  actually 
become  reduced.  The  differences  will  not  disappear,  of 
course,  because  of  the  ineradicable  nature  of  early  influ- 
ences. In  a  group  from  a  small  city  such  as  was  employed 
by  Pressey,  for  example,  the  schooling  of  all  the  children 
was  probably  very  similar  and  hence  there  was  no  in- 
crease in  intellectual  differences  with  age  among  the  oc- 
cupational groups.  The  large  Amount .of  overlapping  of 
scores  in  the  various  groups  is  also  to  be  expected  from 
tEe  equalizing  influence  of  universal  education.  Further- 
more, occupational  classes  are  not  a  sufficient  index  of 
general  cultural  level,  and  this  too  may  account  for 
the  marked  overlapping  found.  From  every  angle,  then, 
the  data  seem  to  be  consistent  with  an  environmental 

No  discussion  of  occupational  intelligence  is  complete 
without  mention  of  the  vast  array  of  data  on  this  subject 
collected  in  the  course  of  the  army  testing.  These  data 
have  subsequently  been  worked  over  to  show  average 
and  range  of  scores  obtained  by  men  in  each  of  96  major 
occupations  (cf.  7).  The  differences  in  average  score  are 
large  when  extreme  groups  are  compared,  but  the  over- 
lapping of  individual  scores  is  pronounced  throughout. 
The  influence  of  specific  environmental  stimulation  is 
again  apparent.  Thus  the  Army  Alpha,  being  predom- 
inantly verbal  in  content,  would  be  expected  to  place 
at  an  advantage  those  individuals  who  have  been  engaged 
in  office  work,  even  if  this  be  of  a  simple  nature.  Such 
has  actually  been  found  to  be  the  case.  Routine  clerical 
workers,  for  example,  obtained  higher  average  scores  (96, 


91)  than  highly  skilled  mechanics  (74,  66).  It  would  seem, 
then,  that  investigations  of  occupational  intelligence, 
whether  conducted  on  adults  or  children,  serve  but  to 
illustrate  the  dependence  of  intelligence  upon  the  psy- 
chological milieu  in  which  the  individual  develops. 


1.  Conradi,  E.     "Song  and  Call-Notes  of  English  Sparrows 
When  Reared  by  Canaries,"  Amer.  J.  PsychoL,  1905,  16, 

2.  Craig,  W.    "The  Stimulation  and  Inhibition  of  Ovulation 
in  Birds  and  Mammals,"  /.  An.  Beh.,  1913,  3,  215-221. 

3.  .    "Male  Doves  Reared  in  Isolation,"  /.  An.  Beh.,  1914, 

4,    121-133. 

4.  Duff,  J.  F.,  and  Thomson,  G.  H.     "The  Social  and  Geo- 
graphical Distribution  of  Intelligence  in  Northumberland," 
Brit.  J.  PsychoL,  1923,  14,  192-198. 

5.  Foley,  J.  P.,  Jr.     "First  Year  Development  of  a  Rhesus 
Monkey  (Macaca  mulatto)  Reared  in  Isolation,"  /.  Genet. 
PsychoL,  1934,  45,  39-i<>5- 

6.  .     "Second  Year  Development  of  a  Rhesus  Monkey 

(Macaca   mulatto)    Reared   in   Isolation    during  the   First 
Eighteen  Months,"  /.  Genet.  PsychoL,  1935,  47,  73-97. 

7.  Fryer,  D.     "Occupational  Intelligence  Standards,"  School 
and  Society,  1922,  16,  273-277. 

8.  Goddard,  H.  H.  Feeblemindedness.  N.  Y. :  Macmillan,  1914. 
Pp.  599. 

9.  Goodenough,  F.  L.     "The  Relation  of  the  Intelligence  of 
Preschool  Children  to  the  Occupation  of  Their  Fathers," 
Amer.  J.  PsychoL,  1928,  40,  284-294. 

10.  Gordon,  H.     Mental  and  Scholastic  Tests  among  Retarded 
Children.  London:  Board  of  Educ.,  Educ.  Pamphlet  No.  44, 

11.  Haggerty,  M.  E.,  and  Nash,  H.  B.    "Mental  Capacity  of 
Children   and   Parental   Occupation,"    /.   Educ.    PsychoL, 
1924,  15,  SS9-S72. 

12.  Hoge,  M.  A.    "The  Influence  of  Temperature  on  the  De- 


velopment  of  a  Mendelian   Character/'  /.  Exper.   Zoo/., 
1915,  i8,  241-285. 

13.  Itard,  J.  M.  G.    The  Wild  Boy  of  Aveyron  (transl.  by  G.  and 
M.  Humphrey).  N,  Y.:  Century,  1932.  Pp.  104. 

14.  Jennings,  H.  S.     The  Biological  Basis  of  Human  Nature. 
N.  Y.:  Norton,  1930.   Pp.  384. 

15.  Kellogg,  W.  N.    "A  Further  Note  on  the  'Wolf  Children' 
of  India,"  Amer.  J.  PsychoL,  1934,  46,  149-150. 

1 6.  Kellogg,  W.  N.,  and  L.  A.    The  Ape  and  the  Child.    N.  Y.: 
Whittlesey  House,  McGraw-Hill,  1933.    Pp.  341. 

17.  McDonald,  H.    "The  Social  Distribution  of  Intelligence  in 
the  Isle  of  Wight,"  Brit.  J.  PsychoL,  1925,  16,  123-129. 

1 8.  Morgan,  T.  H.,  Bridges,  C.  B.,  and  Sturtevant,  A.  H.    The 
Genetics  of  Drosophila.   Hague:  Nijhoff,  1925.    Pp.  262. 

19.  Pressey,  L.  W.     "The  Influence  of  Inadequate  Schooling 
and  Poor  Environment  upon  Results  with  Tests  of  Intelli- 
gence,"  J.  AppL  PsychoL,  4,  1920,  91-96. 

20.  Pressey,  S.  L.,  and  Ralston,  R.    "The  Relation  of  Occupa- 
tion to  Intelligence  as  It  Appears  in  the  School  Children  of 
a  Community,"   /.  AppL  PsychoL,  1919,  3,  366-373. 

21.  Sanders,  B.  S.    Environment  and  Growth.    Baltimore:  War- 
wick and  York,  1934.   Pp.  375. 

22.  Scott,  W.  E.  D.    "Data  on  Song  in  Birds,"  Science,  1901, 
N.  S.  14,  522-526. 

23.  Squires,  P.  C.   "Wolf  Children  of  India,"  Amer.  J.  PsychoL, 
1927,38,  313-315- 

24.  Stockard,  C.  R.    The  Physical  Basis  of  Personality.    N.  Y.: 
Norton,  1931.  Pp.  320. 

25.  Stratton,  G.  M.    "Jungle  Children,"  PsychoL  Bull.,  1934, 
3  i,S96-597. 

26.  Tredgold,  A.  F.    Mental  Deficiency.    N.  Y,:  Wood,  1929. 
Pp.  535- 


The  interpretation  of  family  resemblances  is  compli- 
cated by  the  fact  that  close  relatives  generally  live  to- 
g"etEer.  The  environment  of  individuals  within  a  single 
home  is  certainly  more  similar  than  in  any  other  situation 
outside  of  an  experimental  set-up.  As  a  result,  the  two 
classes  of  factors,  hereditary  and  environmental,  operate 
simultaneously  to  produce  greater  likeness  within  the 
ordinary  family  than  is  found  among  individuals  chosen 
at  random.  The  closer  the  hereditary  relationship,  fur- 
thermore, the  greater  the  environmental  proximity.  Thus 
parents  and  children,  and  brothers  and  sisters  usually  live 
in  the  same  home,  whereas  more  distant  relatives,  such 
as  uncles  and  nephews,  or  cousins,  come  into  less  frequent 
contact.  Not  only  are  related  individuals  exposed  to 
common  environmental  stimulation  because  of  similarity 
of  living  conditions,  but  they  also  constitute  in  part  each 
other's  environment  and  are  rendered  more  alike  by  this 
mutual  interaction.  It^aa^jW^seem  that  family  groupings 
offer  an  excelleirt^j^^  qFTnvlron- 

jdevelopjxient  ^ 


Curiously  enough,  however,  family  resemblances  are 
commonly  attributed  directly  to  the  operation  of  heredity. 
The  child  is  described  as  having  his  father's  business 
acumen,  his  aunt's  musical  talent,  "taking  after"  his 
grandfather  in  obstinacy,  and  perhaps  inheriting  a  keen 
sense  of  humor  from  an  Irish  grandmother  on  his  father's 
side!  The  successful  son  of  an  eminent  family  attributes 



his  accomplishments  to  the  fact  that  he  is  well-born.  A 
lecturer's  vigor  and  zeal  are  explained  by  his  coming  from 
pioneer  stock.  A  boy's  ingenuity  with  mechanical  toys 
is  regarded  as  only  natural  when  one  finds  that  he  is 
descended  from  a  "long  line"  of  boat  builders  and  in- 
ventors. Nor  is  this  type  of  interpretation  limited  to 
popular  slip-shod  thinking  and  everyday  conversation. 
Many  otherwise  accurate  and  well-conducted  scientific 
investigations  on  family  resemblances  contain  the  same 
logical  fallacy. 

The  two  major  methods  employed  in  the  study  of  family 
similarities  and  differences  are  family  history  and  cor- 
relation. The  former  method  has  been  applied  chiefly  by 
eugenicists.  Geneologies  are  traced  and  elaborate  pedigree 
charts  drawn  up  for  families  outstanding  either  for  their 
deficiencies  or  for  their  talents.  The  correlation  studies 
usually  deal  with  the  scores  of  relatives  on  standardized 
tests.  Parents  and  children,  siblings,  and  twins  have  been 
compared  by  this  method.  The  correlation  coefficient  1 
furnishes  a  convenient  numerical  index  of  the  degree  of 
correspondence  between  the  scores  of  any  such  groups. 

It  is  of  course  impossible  to  determine  directly  by  either 
ofjthese  methods  what  is  the  relative  contribution  of  hered- 
itaryjDr  environmental  factors  in  producing,  the  xthtained 
similarities.  Both  methods  are  at  best  descriptive  and 
serve  only  to  discover  more  or  less  objectively  the  degree 

1  The  correlation  coefficient  (r)  is  an  index  of  relationship  between  two  sets  of 
measures.  It  varies  from  +-LOO,  a  perfect  positive  correlation,  through  o,  to 
—  i.oo,  a  perfect  negative  or  inverse  correlation.  A  +1.00  correlation  means 
that  the  highest  score  in  the  one  set  of  measures  is  paired  off  with  the  highest 
in  the  other  set,  the  second  highest  in  the  first  with  the  second  highest  in  the 
second,  etc.  A  —  i.oo  correlation  indicates  that  the  highest  score  in  one  group 
corresponds  to  the  lowest  in  the  other,  a  similar,  perfect  reversal  occurring 
throughout  the  distribution.  A  zero  correlation  indicates  no  relation  at  all,  or 
the  sort  of  arrangement  which  would  result  if  the  scores  in  the  two  sets  were  to 
be  shuffled  and  paired  off  at  random. 


of  familial  resemblance  existing  under  present-day  con- 
ditions of  living.  Only  the  experimental  approach  could 
yield  a  conclusive  solution  to  this  problem.  Thus  if  a 
child  of  known  parentage  were  isolated  from  its  family 
immediately  after  birth  and  brought  up  under  rigidly  con- 
trolled conditions,  many  of  the  questions  on  heredity  and 
environment  might  be  answered.  In  such  an  experiment, 
it  would  also  be  necessary  to  exert  some  control  over  pre- 
natal environment,  such  as  by  proper  care  and  diet  of  the 
mother.  Unfortunately,  popular  sentiment  and  social 
custom  have  stood  in  the  way  of  any  sufficiently  extensive 
experimentation  along  these  lines.  An_ajp£roximation  to 
this  set-up  is,,,,,  however,  afforded  by  the  study  of  foster 
children.  The  earlier  the  child  is  adopted,  the  more  nearly 
does  this  situation  resemble  the  experimental  situation 
described  above.  An  excellent  opportunity  for  the  analysis 
of  hereditary  and  environmental  factors  is  also  furnished 
by  identical  twins  who  have  been  reared  apart  from  an 
early  age,  although  the  number  of  such  cases  is  necessarily 

Because  of  their  more  direct  bearing  upon  the  heredity- 
environment  problem,  all  studies  on  twins  and  on  foster 
children  have  been  reserved  for  a  detailed  treatment  in 
the  next  chapter.  The  present  chapter  will  deal  exclusively 
with  the  more  common  and  general  sort  of  family  rela- 
tionships, including  parents  and  children,  siblings,  and 
more  remote  relatives  or  ancestors. 


Prompted  by  his  interest  in  eugenics  and  the  control  of 
human  evolution,  and  by  his  desire  to  unearth  all  possible 
data  bearing  on  the  laws  of  human  inheritance, 

This  method  was  employed  by  Galton  in  his  extensive 


investigations  on  the  inheritance  of  genius,  the  results  of 
which  were  brought  together  in  1869  in  a  book  entitled 
Hereditary  Genius.  Galtori's  approach  was  distinctly 
heredfitarian,  as  is  illustrated  by  the  following  summary 
of  the  aim  of  his  investigation:  "I  propose  to  show  in 
this  book  that  a  man's  natural  abilities  are  derived  by 
inheritance,  under  exactly  the  same  limitations  as  are  the 
form  and  physical  features  of  the  whole  organic  world" 

(Si  P-  *); 
In  this  investigation,  data  were  collected  on  997  eminent 

men  in  300  families.  In  order  to  facilitate  the  tracing  of 
family  histories  and  the  location  of  descendants  and  other 
relatives,  the  study  was  limited  to  eminent  men  who  were 
either  English  or  well-known  in  England.  The  informa- 
tion was  obtained  from  biographical  collections  or  through 
direct  inquiry  among  relatives  and  acquaintances  of  the 
men  themselves.  Galton  defined  as  follows  the  degree  of 
eminence  necessary  for  inclusion  in  his  survey:  "When  I 
speak  of  an  eminent  man,  I  mean  one  who  has  achieved 
a  position  that  is  attained  by  only  250,  persons  in  each 
million  of  men,  or  by  one  person  in  each  4000"  (5,  p.  9). 
The  classes  of  men  in  Galton'  s  survey  comprised  English 
judges,1  statesmen,  commanders,  literary  men,  scientists, 
poets,  artists  (musicians  and  painters),  and  protestant 
divines,  the  last  including  men  who  had  achieved  fame 
through  some  phase  of  religious  activity,  such  as  theolog- 
ical scholars,  administrators,  religious  leaders,  martyrs, 

man  was  taken 

kinships  were  expressed  in 
Following  the  nameTof  each  of  these  men, 
Galton  appended  a  list  of  famQus  relatives  together  with 
the  major  field  in  which  each  had  achieved  distinction. 

1  The  only  category  limited  exclusively  to  England. 



Whenever  more  complete  information  was  available,  these 
data  were  presented  in  the  form  of  a  family  chart,  a  device 
which  has  gained  in  popularity  among  present-day  eugen- 
icists.  As  a  final  summary  of  his  findings,  Galton  computed 
the, percentage  of  eminent  men  in  each  degree  of  kinship 
to  the  most  eminent  man  of  the  family,  the  latter  still 
SHZ1B£JLS,,  ^e  P°int  °f  reference.  These  percentages  are 
given  in  Table  III  for  each  class  of  "eminence"  sepa- 
rately, as  well  as  for  all  classes  combined.  It  should  be 
noted  that  the  eminent  relatives  within  any  class  have 
not  necessarily  achieved  distinction  in  that  particular 
class;  thus  the  famous  kinsmen  of  a  statesman  may  in~ 


(After  Galton,  5,  p,  308) 

Nature  of  Kinship  * 




CO    § 







































































,    4 











,  *4 


2  * 

















,  .4 


First  cousin 



















'  10* 









:  o 


All  more  remote 




IS  , 






*  No  female  relatives  are  included  in  these  summary  figures,  although  the 
names  and  achievements  of  such  relatives  are  given  in  the  specific  family  his- 
tories, s  ...  '  " 


elude  scientists,  artists,  divines,  etc.  The  classification  is 
based  solely  on  the  field  of  activity  of  the  "most  eminent'9 
man  in  the  family,  around  whom  the  data  are  organized. 

These  figures  suggest  quite  strongly  that  eminence  tends 
to  fun  In  families.  Not  only  are  the  percentages  much 
greater  than  is  expected  by  chance  and  quite  consistent 
from  class  to  class,  but  they  also  show  a  definite  decrease 
in  the  frequency  of  eminent  relatives  as  the  degree  of  re- 
lationship becomes  more  remote.  It  is  quite  a  different 
matter,  however,  to  conclude  that  genius  is  Inherited. 
Galton,  to  be  sure,  recognized  the  difficulties  in  the  way 
of  such  a  conclusion  and  attempted  a  systematic  analysis 
of  them.  To  the  question  of  whether  reputation  is  a  fair 
test  of  ability,  he  answers  in  the  affirmative.  He  argues 
that  reputation  or  eminence,  as  the  criterion  is  employed 
in  his  survey,  is  "the  opinion  of  contemporaries,  revised 
by  posterity — the  favorable  result  of  a  critical  analysis 
of  each  man's  character,  by  many  biographers"  (5,  p.  33), 
and  hence  not  an  accidental  rise  to  short-lived  notoriety. 
Natural  ability  he  defines  quite  circularly  as  "those  qual- 
ities of  intellect  and  disposition,  which  urge  and  qualify  a 
nan  to  perform  acts  that  lead  to  reputation"  (5,  p.  33). 

Although  admitting  the  influence  of  training,  surround- 
ings, and  opportunities,  Galton  minimizes  the  part  which 
they  play  in  the  attainment  of  eminence.  He  constantly 
holds  up  to  the  reader  the  heroic  picture  of  genius  triumph- 
ing over  obstacles.  By  definition,  genius  means  to  him 
"a  nature  which,  when  left  to  itself,  will,  urged  by  an 
inherent  stimulus,  climb  the  path  that  leads  to  emi- 
nence, and  has  strength  to  reach  the  summit — one  which, 
if  hindered  or  thwarted,  will  fret  and  strive  until  the 
hindrance  is  overcome,  and  it  is  again  free  to  follow  its 
labour-loving  instinct"  (5,  pp.  33-34).  He  concludes  that 
"  It  is  almost  a  contradiction  in  terms,  to  doubt  that  such 


men  will  generally  become  eminent/'  and  adds  that  "  there 
is  plenty  of  evidence  in  this  volume  to  show  that  few  have 
won  high  reputations  without  possessing  these  peculiar 
gifts"  (5,  p.  34).  This  is  true  enough,  but  it  remains  to  be 
proved  that  such  "gifts"  as  the  impulse  to  climb,  the 
strength  to  reach  the  summit,  and  the  love  of  labor  are 
themselves  independent  of  environment.  Unfortunately, 
the  optimistic  picture  painted,  by  Galton  is  not  borne  out 
t>y  observations  of  everyday  life;  and  in  the  absence  of 
experimental  proof,  it  is  impossible  to  accept  Galton's 
interpretations  of  his  findings.1 


The  family  history  method  has  also  been  widely  em- 
ployed in  the  effort  to  analyze  the  causes  of  intellectual 
defect,  crime,  pauperism,  and  similar  conditions.  By  this 
method,  a  number  of  families  have  been  discovered  which 
present  an  overwhelming  array  of  socially  inadequate  per- 
sons over  several  generations.  The  same  general  tech- 
niques are  used  in  tracing  the  history  of  these  families  as 
in  the  study  of  eminent  groups.  Living  relatives  or  de- 
scendants are  visited  and  observed,  residents  of  the  vicin- 
ity are  interviewed,  and  certificates  of  marriage  and  birth, 
and  similar  public  records  are  examined  whenever  avail- 
able. These  families  are  usually  found  in  rural  districts 
in  many  parts  of  the  country,  often  inhabiting  the  same 
crude  huts  built  by  their  ancestors  many  generations  ago. 
They  frequently  interbreed,  are  quite  prolific,  and  even- 
tually come  to  constitute  their  own  community,  shunned 
and  ridiculed  by  their  neighbors. 

The  earliest  published  pedigree  of  such  a  "degenerate" 

1  Other  studies  on  eminent  families  will  be  found  in  Ch.  XIII  on  Genius. 
Galton's  study  is  here  reported  only  as  an  illustration  of  the  family  history 








Key  to  Symbols 

Feebleminded  Man    % 
Feebleminded  Woman 
A/    Normal 

i  MARRIED  r 



(A)  Normal  Woman       v 

(After  Goddard,  7,  p.  36.) 



dard  constantly  emphasizes  the  role  of  heredity.  Having 
laid  great  stress  upon  the  fact  that  the  two  groups  were 
branches  of  the  same  family,  furnishing,  "as  it  were,  a 
natural  experiment  with  a  normal  branch  with  which  to 
compare  our  defective  side,"  he  states  that  "from  this 
comparison,  the  conclusion  is  inevitable  that  all  this  de- 
generacy- has  come  as  the  result  of  the  defective  mentality 
and  bad  blood  having  been  brought  into  the  normal 
family  of  good  blood"  (7,  pp.  68-69).  It  seems  rather 
curious  that  the  common  descent  of  the  two  branches 
from  Martin  Kallikak  should  be  regarded  as  strengthen- 
ing an  hereditary  interpretation  of  the  differences  be- 
tween them.  The  environments  of  the  two  groups  were 
not  in  any  way  equated  by  this  common  ancestry  and 
in  fact,  it  is  evident  that  the  members  of  the  two 
branches  were  reared  under  widely  differing  conditions. 
Common  descent  from  Martin  Kallikak  simply  made 
the  heredity  of  the  two  groups  more  alike  than  would 
be  the  case  in  the  comparison  of  two  distinct  families 
and  thus  seems  to  strengthen  an  environmental  explana- 
tion of  tKe  differences,  if  it  contributes  anything  at 

A  more  crucial  test  would  have  been  available  if  the 
legitimate  offspring  of  Martin  and  his  well-born  wife  had 
been  exchanged  at  birth  with  those  of  the  feebleminded 
woman.  It  would  then  have  been  very  illuminating  to 
ascertain  the  relative  percentage  of  feeblemindedness  and 
other  defects  in  the  "normal"  and  "degenerate"  stock. 
The  practical  obstacles  in  the  way  of  such  a  procedure  in 
no  way  strengthen  the  validity  of  conclusions  drawn  from 
an  inadequately  controlled  situation. 

Many  equally  "degenerate"  families  have  been  subse- 
quently investigated  by  psychologists,  sociologists,  or  eu- 
genicists.  The  research  staff  of  the  Eugenics  Record 


Office  1  conducts  such  surveys  as  one  phase  of  their  regular 
work.  Among  the  groups  they  have  studied  are  the 
Hill  Folk,  the  Nam  family,  and  the  W  family  of  Indiana, 
all  presenting  the  same  picture  of  degeneracy,  mental 
defect,  disease,  and  social  incompetence  through  suc- 
cessive generations.  Surveys  of  eminent  families  have 
likewise  been  sponsored  by  the  Eugenics  Record  Office. 
Specific  lines  of  achievement,  such  as  scholarly  pursuits 
or  boat  designing,  have  been  traced  from  generation  to 
generation  in  the  attempt  to  show  that  such  talents 
are  transmitted  through  heredity.  Although  offering  much 
interesting  material,  such  studies  cannot  yield  any  data 
on  the  heredity-environment  question;  the  opportunities 
for  environmental  transmission  of  such  family  qualities 
are  too  obvious  to  overlook  or  dismiss. 


the  correlation  technique,  although  more 
accurate  and  quantitative,  does  not  eliminate  the  essen- 
tiaL.jdrfficuIfy  inherent  .  in  the  family  history  .  ^method, 
naively.,  the  confusion  of  hereditary  and  environmental 
CQiLtribiitioi\s.  Pearson  (14)  was  among  the  first  to  apply 
correlation  analysis  to  parent-child  resemblances.  Con- 
tinuing a  line  'Bf  research  initiated  by  Galton  (6),  he  col- 
lected measures  on  A  ^^  garents  and  offspring  in  physical 
traits  such  as  stature,  arm  span,  and  forearm  Je&gth. 
ThE_|)ajeat--cliild  correlations  in  these  traits'  averaged 
about  .52.  The  similarity  of  this  correlation  to  those 
oBlamed  for  bodily  characteristics  of  many  animal  forms 
led  Pearson  and  others  to  suggest  thal^his^figure  indi- 

of  ^here.dit^x  J^c,tor&  -to.  -the!  de- 
Family  resemblance  in 

1  Eugenics  Record  Office,  Cold  Spring  Harbor,  Long  Island,  N.  Y.    Cf.,  e.g., 
references  (i)  and  (4). 


such  traits  Is  doubtlessly  owing  in  large  part  to  the  infly- 
ence  of  common  heredity,  "although.,  the,  xole  ,of .  similar 
environment,  especially  in  the  prenatal  stage,  cannot  be 

A  crude  but  frequently  quoted  early  study  on  parental 
resemblance  in  psychological  traits  is  that  of  Woods  (19) 
on  royal  families.  Because  of  the  large  amount  of  inbreed- 
ing among  royal  houses,  such  a  group  seemed  to  offer  an 
especially  good  opportunity  for  the  analysis  of  family 
likenesses.  In  addition,  genealogies  could  be  traced  more 
easily  in  royalty,  and  information  regarding  the  char- 
acters and  abilities  of  royal  personages  was  relatively 
accessible.  Each  individual  investigated  was  assigned  a 
rating  from  I  to  10,  representing  different  degrees  of  in- 
tellect from  feeblemindedness  to  genius.  By  a  similar 
procedure,  each  was  rated  for  moral  qualities.  The  rat- 
ings were  based  upon  available  historical  and  biographical 
material  and  are  therefore  subject  to  all  the  errors  in- 
herent in  such  data,  plus  a  possible  error  of  judgment 
introduced  by  Woods'  own  evaluation  of  the  facts. 

In  the  entire  survey,  data  on  several  thousand  persons 
from  various  European  royal  houses  were  employed.  From 
the  available  information,  it  proved  possible  to  assign 
ratings  on  intellectual  quality  to  a  total  of  671  cases  and 
on  moral  traits  to  608.  The  correlations  between  the 
ratings  of  offspring  and  fathers,  grandfathers,  and  great- 
grandfathers, respectively,  are  given  below.1 

Relationship  Intellect  Morals 
Fathers  and  offspring                                      .30  .30 

Grandfathers  and  offspring  .16  .18 

Great-grandfathers  and  offspring  .15 

1  These  correlation  coefficients,  although  utilizing  the  ratings  on  the  lo-point 
scale,  were  computed  by  means  of  a  four-fold  correlation  table.  All  individuals 
receiving  ratings  from  I  to  5  were  classified  as  below  average,  those  between  6 
and  10  as  above  average. 


That  there  exists  a  noticeable  correspondence  and  that 
this  correspondence  is  stronger  the  closer  the  relation- 
ship, seems  clear  from  these  correlations.  Little,  can  be 
concluded,  however,  regarding  the  amount  of  relation- 
ship, since*  the  size  of  a  correlation  coefficient  is  too  largely 
influenced  by  -such.  errors,  of  measurement  as  were  present 
in  these  data. 

More  recently,  scores  obtained  by  parents  and  children 
on  standardized  tests  have  been  correlated,  .Jones  (n) 
adrninistered  intelligence  tests  to  over  300  families  in,  a 
rural  district,  in  New  England.  In  order  to  minimize  the 
factors  of  language  handicap  and  differential  group  tradi- 
tions, communities  were  chosen  in  which  the  population 
was  composed  entirely  of  native-born  stock.  All  children 
between  the  ages  of  3  and  14  were  given  the  Stanford- 
Binet;  older  offspring  and  all  parents  were  tested  with  the 
Army  Alpha.  In  105  families,  it  was  possible  to  test  both 
parents  and  two  or  more  children.  It  was  upon  this 
sampling,  consisting  of  210  parents  and  317  children, 
that  the  parent-child  correlations  were  computed. 

The  following  are  some  of  the  correlations  1  reported 
by  Jones. 

0  n       ,  Son  and 

son  Daughter  r>       7j 

6  Daughter 

Father  .524  .505  ,509 

Mother  .544  .557      "  .548 

The  corrfil&tions-QfrJi^  i.e.,  father 

and,  .spri,  or  mother  aiuL-daughter,  ara,~aaL,,,sigi]diip,antly 
higher  %than  those,  .between  unlike-sex-  pareat  ..and.  child. 

1  It  is  a.  frequent  practice,  in  studies  on  parent-child  resemblance,  to  compute 
a  "midparent  "-child  correlation,  in  which  the  average  score  of  the  two  parents 
is  correlated  with  the  child's  score  in  order  to  furnish  a  final  summary  measure 
of  relationship.  Such  a  correlation  is,  however,  misleading,  because  in  the  process 
of  averaging  scores  a  certain  amount  of  individual  variation  is  eliminated  and  as 
a  result  the  correlation  may  be  too  high. 


Thusjthere  seemed  to  bejio  basis,  to  the  belief  that  boys 
tend  to  resemble  more  closely  their  fathers,  and  girls 
their  mothers  in  intellectual  level.  There  was3  however,  a 
small  but  consistent  tendency  for  the  scores  of  children 
of  either  sex  to  correlate  more  highly  with  the  mother's 
score  than  with  that  of  the  father.  This  might  be  at- 
tributed to  the  closer  contact  which  the  mother  has  with 
the  children. 

In  a  similar  study  by  \Yillqughby  (17)  eleven  tests  were 
administered  to  about  141  children,  100  mothers,  and 
90  fathers,  all  of  whom  lived  in  the  vicinity  of  Palo  Alto, 
California.  The  tests  ranged  from  highly  verbal  to  highly 
non-verbal,  and  were  selected  from  various  current  scales 
for  the  measurement  of  intelligence  and  scholastic  achieve- 
ment. The  average  of  the  parent-child  correlations  in  all 
of  the  tests  was  .35.  There  was  no  consistent  tendency 
for  the  correlations  to  be  higher  with  either  parent.  A 
comparison  of  the  correlations  on  each  test,  however, 
reveals  large  discrepancies.  The  average  parental  corre- 
lations 1  on  different  tests  ranged  from  .17  to,  .48.  When 
the  tests  were  classified  into  verbal  and  non-verbal,  the 
average  parental  correlation  was  .39  for  the  former  and  .30 
for  the  latter.  This  difference  is  also  borne  out  by  a  com- 
parison of  individual  tests.  Thus  the  three  highest  cor- 
relations., ^ere.  obtained  with  rdat|y,ely  abstract,  tests  in 
which  learning  and  past  experience  predominate,  viz., 
history-literature  information  (.48),  number-series  comple- 
tion (.42),  and  analogies  (.41);  the  three  yielding  the 
lowest  correlations  were  checking  similarities  (.17),  geo- 
metric forms  test  from  Army  Beta  (.28),  and  symbol- 
digit  code  learning  test,  also  from  Army  Beta  (.30). 

The  results  of  ,W^illoughby  again  suggest  the 

1  Average  of  the  following  separate  correlations:  mother-daughter,  mother-soa, 
father-daughter,  and  father-son. 


by  environment  and  stimulation  in  determining  the  degree 
of  familial  resemblance.  Performance  on  verbal  tests  is 
more  susceptible  to  training  and  home  conditions  than  is 
performance  on  the  non-verbal,  and  it  is  the  former  which 
yielded  the  highest  parental  correlations.  In  Willoughby's 
study,  furthermore,  the  correlations  run  lower  than  those 
of  Jones.  This,  too,  would  be  expected  from  the  nature  of 
the  tests,  since  the  latter  used  only  intelligence  tests  of  a 
highly  verbal  type.  The  fact  that  Jones  correlated  total 
scores  on  composite  tests  made  up  of  many  parts,  whereas 
Willoughby  employed  scores  on  single  and  fairly  homo- 
geneous tests,  may  also  account  for  the  higher  correla- 
tions found  by  Jones.  The  more  complex  the  measures 
correlated,  the  greater  are  the  chances  .that  they..^ill 
have  common  elements  and  the  higher  will  their  inter- 
correlations  tend  to  be. 

Certain  tentative  conclusions  are  suggested  by  an  anal- 
ysis of  the  data  on  parental  resemblance.  In  the  first 
place,  parents  and  children  exhibit  a  distinct  similarity 
in  physical  traits,  which  r&ay  b<e  expressed  by  a  correla- 
tion in  the  neighborhood  of  .50.  Approximately  the  sai^e 
degree  of  correspondence  is  found  on  most  .common  in- 
telligence tests  of  t  the  verbal  type,  \yhen,  however,  Qpm- 
parisons  are  made  on  more  homogeneous  .tests^Qi^on 
tests  which  are  less  susceptible  to  common  home  environ- 

'         "'    '"      •  A  '  ,,,,„<»?<    »P  «-»•.» 

ment,  the  correlations  are  much  lower.  There  is  a  sug- 
gestion of  a  closer  resemblance  to  the  mother  than  to  the 
father  in  intelligence  test  performance,  although  the  data 
on  this  point  are  inconclusive. 


Investigations  on  the  resemblance  of  siblings  have  been 
much  more  numerous  than  those  on  parent-child  similar- 
ity. As  Pearson  pointed  out,  the  comparison  of  siblings, 


especially  when  both  are  in  school,  does  not  present  the 
practical  difficulties  met  in  the  testing  or  rating  of  parents. 
In  addition  to  extensive  comparisons  of  siblings  in  physical 
traits  such  as  eye-color.,  hair-color  and  curliness,  health, 
and  head  dimensions,  in  which  about  the  same  degree  of 
relationship  was  found  as  in  the  parent-child  studies, 
Pearson  (13)  undertook  an  investigation  of, .sibling  re- 
semblances in  mental  traits.  The  data  of  the  latter  study 
consisted  exclusively  of  teachers'  ratings  on  schoolchildren. 
The  measures  used  are  necessarily  crude  and  may  be 
biased  by  the  teachers5  own  reaction  to  family  relation- 
ships; thus,  two  brothers  might  be  rated  more  nearly 
alike  because  the  teacher  knows  that  they  come  from  the 
same  family.  The  correlations  are  based  on  large  sam- 
plings, however,  the  number  of  cases  varying  from  554 
to  2152  for  different  traits.  Below  are  the  correlations 
found  by  Pearson  between  pairs  of  brothers,  pairs  of 
sisters,  and  mixed  pairs,  consisting  of  a  brother  and  a 

Trait  Brothers  Sisters         Mixed  Pairs 

Vivacity  47  .64  .63 

Self-assertiveness  .53  .49  .51 

Popularity  .50  .56  .48 

Conscientiousness  .59  .61  .52 

Temper  .51  43  49 

Ability  46  44  .52 

Handwriting  .53  47  .63 

Introspection  .59  47  44 

Average  .52  .57  .49 

Pearson  himself  gives  a  strongly  hereditarian  interpreta- 
tiorTto  these  findings,  offering  the  opinion  that:  "Jt. js,  the 

stockjtself  wjiich f makes  its^  home^aiv^^ 
cation  is  of  small  service^ul^&lV^ 

gent  rrc^Slmen*y  (13,  p.  159).  It  is  characteristic  of  many 
-  t|ie  influence  of  environment  is  min- 


imized,  to  cite  formal  education  as  the  principal  environ- 
mental force  and  to  overlook  the  broader  and  probably 
more  effective  aspects  of  general  environment.  Since 
Pearson  found  very  similar  sibling  correlations  in  mental 
anct  physical  traits,  and  .since,  he  believed  the  latter  could 
beTIttie  influenced  by  environment,  he  concluded  that 
thelame  must  be  true  of  mental  traits.  Such  a  conclusion 
doesTiot  necessarily  follow.  It  is  itself  based  on  the  as- 
sumption that  mental  and  physical  traits  are  inherited  in 
the  same  manner  and  thus  involves  a  circular  argument. 

Investigations  with  mental  tests  have  all  yielded  posi- 
tive and  significant  sibling  correlations,  although  the 
amount  of  relationship  varies  with  the  particular  test, 
the  age  of  the  subjects,  and  other  conditions.  Gordon  (8) 
found  a  correlation  of  .53  between  the  Stanford-Binet 
I.Q.'s  of  91  pairs  of  siblings  in  an  orphanage  in  California. 
Pintner  (15)  administered  six  simple  tests  to  180  pairs  of 
siblings  and  computed  "mental  indices"  for  each  child  on 
the  basis  of  scores  on  all  the  tests;  the  sibling  correlation 
in  mental  indices  was  .22.  Hart  (9)  correlated  the  I.Q.'s 
of  siblings  in  three  groups  of  schoolchildren,  the  number 
of  sibling  pairs  in  each  group  being  252,  147,  and  219, 
respectively.  The  I.Q/s  were  found  from  scores  on  Army 
Alpha,  National  Intelligence  Test,  and  Stanford-Binet. 
Correlations  of  .447,  .459,  and  .399  were  obtained  in  the 
three  groups. 

In  an  extensive  investigation  of  sibling  resemblance, 
Hildreth  (10)  analyzed  the  LQ.'s  of  three  large  groups  of 
schoolcKildren.  The  first  group  consisted  of  671  public 
schoolchildren  from  300  families  in  Oklahoma  City;  this 
was  a  group  of  average  social  status.  The  second  group 
was  composed  of  523  children  attending  the  Horace  Mann 
School  in  New  York  City,  the  number  of  families  repre- 
sented totaling  241;  the  social  and  cultural  level  of  this 


group  was  definitely  superior.  The  third  was  a  group  of 
inferior  social  status  obtained  from  the  Hebrew  Orphan 
Asylum  in  New  York  City;  it  included  346  children  in 
146  families.  The  number  of  siblings  within  any  single 
family  ranged  from  two  to  five.  The  I.Q.'s  were  based 
on  some  form  of  the  Binet  scales,  about  98%  of  them 
having  been  obtained  with  the  Stanford  Revision. 

The  sibling  correlations  in  each  of  the  three  groups, 
in  the  order  named  above^  were  .629,  .274,  and  ,322, 
respectively.  The  variation  from  group  to  group  is  large, 
especially  since  the  same  test  was  employed.  The  discrep- 
ancies probably  result  chiefly  from  differences  in  hetero- 
geneity^of  tHe^giaups.  TheJHprace  Mann  group,  being 
thFmost  highly  select  and  homogeneous,  gave  the  lowest 
correlation,  whereas  the  Oklahoma  group,  a  more  nearly 
random  sample  than  either  of  the  other  two,  yielded  the 
highest  correlation.  In  general,  the  wider  the  range  of 
individual  differences  within  a  group,  the  higher  will  be 
the  correlations.1  Hildreth  concludes  that  the  sibling 
correlation  in  intelligence  would  "certainly  appear  to  be 
greater  than  .3  and  less  than  .7"  (10,  p.  56),  and  proposes 
"  heredity  rather  than  environment  as  the  cause  of  the 
resemblance  found"  (10,  p.  60),  although  admitting  that 
the  data  relevant  to  the  latter  point  are  fairly  scant. 

Thorndike  (16)  reported  a  correlation  2  of  .60  between 
the  scores  of  about  1200  pairs  of  siblings  on  the  I.E.R. 
Tests  of  Selective  and  Relational  Thinking,  Generaliza- 
tion, and  Organization.3  Since  all  of  the  subjects  were 

1  For  a  fuller  discussion  of  the  effect  of  homogeneity  of  the  sampling  upon  the 
size  of  a  correlation  coefficient,  the  reader  is  referred  to  Ch.  XI. 

2  The  value  given  (.60)  was  estimated  by  Thorndike  after  applying  corrections 
for  selective  factors  and  other  conditions, 

3  A  series  of  tests  of  the  type  usually  found  in  intelligence  scales,  including 
opposites,  analogies,  sentence  completion,  arithmetic  reasoning,  number  series 
completion,  etc.  They  have  an  added  advantage  in  that  the  scores  are  expressed 
in  terms  of  an  equal  unit  scale. 


attending  high  school,  this  group  is  older  and  more  select 
than  those  ordinarily  employed  in  other  sibling  studies. 
In  interpreting  his  findings,  Thorndike  follows  the  argu- 
ment originally  proposed  by  Pearson  (cf.  above).  Thus 
he  concludes  that  the  difference  of  .08  between  the  cor- 
relation of  .60  on  the  LE.R.  test  and  that  of  .52  estab- 
lished by  Pearson  (13)  for  physical  resemblances  in  sib- 
lings may  be  attributed  to  the  equalizing  influence  of 
similar  school  environment.  This  would  only  follow  if  the 
influence  of  heredity  upon  mental  traits  were  known  to  be 
identical  to  its  influence  upon  physical  traits,  and  if  in 
addition  it  could  be  assumed  that  physical  traits  develop 
incomplete  independence  of  environmental  factors. 

Willoughby  (17),  in  his  investigation  of  family  resem- 
blance cited  in  the  preceding  section,  also  obtained  data 
on  sibling  relationships  in  the  eleven  mental  tests  which 
he  administered.  The  subjects  for  this  part  of  the  study 
were  280  siblings,  ranging  in  age  from  seven  up.  The 
average  and  range  of  the  correlation  coefficients  for  the 
different  tests  were  as  follows : 

Average  i 
(j/  Tests) 

Average  r 

Average  r 

Range  in  r 




.18  to  .58 




.26  to  .63 




.24  to  .54 

Unlike-sex  siblings 

Itu  is  noteworthy  that  the  like-sex  siblings  gave  higher 
correlations  thai  tte  unllkersex.  This  difference,  although 
small,  suggests  the  greater  community,  pf  .environment 
furnished,  within  the  family  for  like-sex  siblings.  As  was 
the  case  with  parental  correlations,  the  sibling  correlations 
exhibit  a  pronounced  variation  from  test  to  test,  and  in 
two  of  the  three  sibling  groups,  the  average  correlation  is 
higher  for  the  verbal  than  for  the  non-verbal  tests. 


In  their  extensive  investigations  of  honesty  In  school- 
children, May  and  Hartshorne  (12)  computed  sibling 
correlations  on  four  types  of  deception  tests.  A  total  of 
734  pairs  of  siblings  in  seven  groups  were  measured.  The 
correlations,1  differing  somewhat  with  the  group  and  the 
test,  ranged  from  .208  to  445.  May  and  Hartshorne  min- 
imize the  part  played  by  environment  in  these  sibling 
resemblances,  their  main  argument  resting  on  the  finding 
that  socio-economic  level  of  the  home  did  not  correlate 
very  highly  with  honesty  scores.  Other  less  easily  ob- 
servable aspects  of  home  environment  may,  however,  be 
much  more  significant  than  socio-economic  level  in  the 
development  of  such  character  traits. 

It  would  seem  that  the  correlations  in  test  performance 
oFlsTblings  reared  together,  although  uniformly  positive 
arid"  reliable,  show  a  marked  variation  in  amount.  Among 
the  major  factors  which  determine  this  variation  may  be 
mentioned  the  nature  of  the  test  and  in  particular  its 
relative  dependence  upon  environment  and  training,  the 
homogeneity  of  the  sampling,  and  the  age  of  the  subjects. 
In  regard  to  the  factor  of  age,  it  is  apparent  that  the  older 
the  subjects,  the  longer  will  environmental  influences  have- 
been  operative.  Whether  such  influences  will  have  a 
leveling  or  a  differentiating  effect,  however,  will  depend 
upon  the  specific  circumstances  in  each  case. 


1.  Danielson,  F.  H.,  and  Davenport,  C.  B.     The  Hill  Folk. 
Cold  Spring  Harbor:  Eugenics  Record  Office  Memoir  No.  I, 
1912.   Pp.  56. 

2.  Dugdale,  R.  L.    The  Jukes:  a  Study  in  Crime,  Pauperism, 

1  One  type  of  deceptive  behavior,  mz.,  securing  help  contrary  to  instructions 
on  a  test  taken  at  home,  gave  much  higher  sibling  correlations,  but  this  was 
definitely  affected  by  collusion,  the  siblings  either  helping  one  another  or  en- 
couraging each  other  in  the  use  of  forbidden  aids. 


Disease,  and  Heredity.   N.  Y.:  Putnam,  1910  (ist  ed.  1877). 
Pp.  1 20. 

3.  Estabrook,  A.  H.   The  Jukes  in  1915.  Washington:  Carnegie 
Institution,  1916.    Pp.  85. 

4.  Estabrook,  A.  H.,  and  Davenport,  C.  B.    The  Nam  Family, 
Cold  Spring  Harbor:  Eugenics  Record  Office  Memoir  No.  2, 
1912.  Pp.  85. 

5.  Galton,  F.  Hereditary  Genius:  an  Inquiry  into  Its  Laws  and 
Consequences.  London:  Macmillan,  1914.  Pp.368. 

6.  .   Natural  Inheritance.   London:  Macmillan,  1889.   Pp. 


7.  Goddard,  H.  H.    The  Kallikak  Family:  a  Study  in  the  Hered- 
ity of  Feeblemindedness.    N.  Y.:  Macmillan,  1921  (ist  ed. 
1912).  Pp.  121. 

8.  Gordon,  K.     "Psychological  Tests  of  Orphan  Children," 
J.Del.,  1919,  4, 46-55. 

9.  Hart,  H.    "Correlations  between  Intelligence  Quotients  of 
Siblings,"  Sch.  and  Soc.,  1924,  20,  382. 

10.  Hildreth,  G.  H.    "The  Resemblance  of  Siblings  in  Intelli- 
gence and  Achievement,"  Teachers  College,  Columbia  Univ., 
Contrib.  to  Educ.,  1925,  No.  186.  Pp.  65. 

11.  Jones,  H.  E.    "A  First  Study  of  Parent-Child  Resemblance 
in  Intelligence,"  2fih  Yearbook,  Nat.  Soc.  Stud.  Educ.,  1928, 
Part  I,  61-72. 

12.  May,  M.  A.,  and  Hartshorne,  H.    "Sibling  Resemblance  in 
Deception,"  2jth   Yearbook,  Nat.  Soc.  Stud.  Educ.,   1928, 
Part  II,  161-177. 

13.  Pearson,  K.    "On  the  Laws  of  Inheritance  in  Man:  II.  On 
the  Inheritance  of  the  Mental  and  Moral  Characters  in 
Man,  and  Its  Comparison  with  the  Inheritance  of  Phys- 
ical Characters,"  Biom.,  1904,  3,  131-190. 

14.  Pearson,  K.,  and  Lee,  A.    "On  the  Laws  of  Inheritance  in 
Man:  I.  Inheritance  of  Physical  Characters,"  Biom.,  1903, 
2,  3S7-462. 

15.  Pintner,  R.     "The  Mental  Indices  of  Siblings,"  Psychol 
Rev.,  1918,  25,  252-255. 

16.  Thorndike,  E.  L.,  and  staff.   "The  Resemblance  of  Siblings 


in   Intelligence,"  2fth    Yearbook,  Nat.   Soc.   Stud.   Educ.} 
1928,  Part  1,41-53. 

17.  Willoughby,  R,  R.     "Family  Similarities  in  Mental  Test 
Abilities,"   27th    Yearbook,   Nat.   Soc.    Stud.  Educ.,    1928, 
Part  I,  55-59. 

1 8.  Winship,  A.  E.    Jukes-Edwards:  a  Study  in  Education  and 
Heredity.  Harrisburg,  Pa.:  Myers,  1900.  Pp.88. 

19.  Woods,   F.  A.      Mental  and  Moral  Heredity  in  Royalty. 
N.Y.:  Holt,  1906.  Pp.  3 12. 


In  the  present  chapter,  a  survey  of  psychological 
studies  on  twins  and  foster  children  will  be  presented. 
These  two  types  of  investigation  of  familial  resemblance 
have  been  selected  for  a  more  detailed  treatment  because 
of  the  special  facilities  which  they  offer  for  the  analysis 
of  the  relative  contribution  of  hereditary  and  environ- 
mental factors.  Twins  and  foster  children  may  be  said  to 
represent  the  two  extremes  of  hereditary  similarity.  In 
the  case  of  identical  twins,  heredity  is  completely  alike  for 
the  two  individuals,  so  that  any  differences  between  them 
may_be  attributed  "directly  to  the  operation  of  different 
stimulating  conditions.  Fotfer^  children)  on  the  other 
ha.nd3  bear  no  hereditary  resemblance  to  their  adopted 
parents  or  to  any  other  children  with  whom  they  are 
reared.  These  subjects  therefore,  Reveal  the  contribution 
of  environmental  influences  in  any  similarity  which  they 
may  exhibit  to  their  foster  parents  or  foster  siblings. 

The  observation  and  measurement  of  fraternal  or  non- 
identical  twins  also  offer  a  fruitful  approach  to  this  general 
problem.  Such  twins  are  no  more  alike  than  ordinary 
siblings  in  respect  to  heredity.  They  have,  however,  been 
exposed  to  the  same  prenatal  environment.  Since  they 
are  of  the  same  age,  they  are  also  subjected  to  postnatal 
stimulation  which  is  much  more  similar  than  in  the  case 
of  ordinary  siblings.  They  would  thus  seem  to  offer  a 
sort  of  "environmental  control"  in  the  analysis  of  the 
sibling  resemblances  ordinarily  observed.  Similarly,  the 
cases  of  identical  twins  who  have  lived  apart  from  early 



infancy  may  be  regarded  as  an  "  hereditary  control"  for 
identical  twins  reared  together  in  the  usual  way.  Both 
of  these  groups  have  the  same  degree  of  hereditary  com- 
munity; any  differences  in  amount  of  resemblance  are 
therefore  attributable  directly  to  environmental  varia- 


Twins  have  long  been  acclaimed  as  particularly  good 
material  for  the  analysis  of  behavioral  resemblance.1  The 
arguments  which  have  been  put  forth  in  the  effort  to 
prove  the  major  potency  of  -heredity  on  the  basis  of  twin 
resemblance.,  summarized  under  five  headings. 
A  brief  critical  evaluation  of  each  will  be  given. 

guiXLent  ,is  based  on  the  relative  degree  of 
pf  siblings.  The  closer  resem- 
blance of  the  former  cannot,  however,  be  attributed  in- 
discriminately to  their  more  similar  heredity.  It  is  true 
that,  in  respect  to  general  home  background  and  school- 
ing, the  environment  of  siblings  is  about  as  similar  as 
that  of  twins.  But  the  attitude  of  the  parents  towards  an 
older  and  a  younger  sibling,  the  attitude  of  the  children 
towards  each  other,  and  the  specific  events  and  vicissi- 
tudes occurring  at  any  one  life  period  for  each  child  will 
differ  far  more  for  siblings  than  for  twins.  Conditions  of 
this  sort,  operating  differentially  from  a  very  early  age, 
may  be  even  more  important  than  economic  level  or 
formal  schooling  in  the  development  of  the  child's  men- 
tality and  character. 

Ajsecpnd  type  of  cpmparison  is  that 

I*  *s  °ften  Argued  that  if  environment 
were  a  significant  factor  in  the  resemblance  of  twins, 
then  this  resemblance  should  be  closer  in  the  older  twins 

1  For  a  discussion  of  the  biology  and  physiology  of  twins,  cf.  Newman  (10,  n).  • 


who  have  been  exposed  for  a  longer  period  to  the  equaliz- 
ing environmental  influences.  As  has  been  repeatedly 
pointed  out  above,  however,  it  is  impossible  to  generalize 
regarding  the  particular  direction  which  environmental 
influences  will  take.  Thus  it  seems  very  probable  that  in 
the  case  of  twins,  early  environment  may  have  a  strongly 
equalizing  eifect,  whereas  later  environment  may  produce 
differentiation.  Prenatal  environment,  to  be  sure,  is 
very  similar  in  such  cases.  In  infancy  and  early  child- 
hood, the  twins  are  usually  treated  in  much  the  same  way. 
As  the  children  grow  older,  however,  any  differences  be- 
tween them  tend  to  be  noted  and  emphasized  for  practical 
reasons,  and  the  twin  begins  to  learn  that  he  must  lead 
his  own  life  and  develop  as  an  individual  if  he  is  to  be 
well-adjusted.  It  would  seem  that  a  drop,  rather  than  a 
rise,  in  correlation  with  age  is  to  be  expected  on  environ- 
mental grounds. 

In  some  of  the  more  recent  studies  the  attempt 
has  been  made  to  classify  twins  into  identical  and  non- 
identical,  or  fraternal.  The  comparison  of  these  two  types 
of  twins  in  mental  traits  has  furnished  a  third  approach 
to  the  analysis  of  hereditary  and  environmental  factors. 
It  will  be  recalled  that  identical  twins  develop  from  a 
single  fertilized  ovum  and  therefore  have  identical  hered- 
ity, whereas  fraternal  twins  are  no  more  similar  than 
ordinary  siblings  in  their  hereditary  constitution,  although 
they  develop  simultaneously.  It  would  seem  that  the 
difference  in  degree  of  resemblance  of  identical  and  non- 
identical  twins  could  be  attributed  directly  to  heredity. 
The  environment  of  fraternal  twins  is  doubtlessly  more 
uniform  than  that  of  siblings  born  at  different  times. 
But  is  it  necessarily  as  homogeneous  as  that  of  identical 
twins?  It  must  be  remembered  that  fraternal  twins  are 
often  of  unlike  sex,  and  the  differential  treatment  of  the 


two  sexes  may  enter  in  to  render  their  environment  dis- 
similar. This  may  prove  a  very  potent  differentiating 
influence  in  the  development  of  the  two  children.  The 
physical  differences  themselves  which  characterize  the 
fraternal  as  contrasted  with  the  identical  twins  may 
further  encourage  differential  reactions  towards  the 
former,  especially  during  the  important  earlier  ages  when 
the  identical  twins  are  probably  still  being  confused  by 
their  associates. 

A  fourth  analysis  is  based  upon  the  relative  degree  of 
twin  resemblance  in  traits  differing  in  their  susceptibility 
to  training.  Such  a  comparison  is  very  illuminating, 
provided  the  traits  are  not  simply  classified  in  respect  to 
their  dependence  upon  school  training.  Home  environ- 
ment is  far  more  significant  than  formal  schooling  in  de- 
termining twin  or  any  other  family  resemblance.  In  those 
tests  which  depend  primarily  upon  school  training,  we 
should  expect  unrelated  children  to  resemble  each  other 
nearly  as  much  as  related  children.  This  is  especially 
true  in  studies  in  which  all  of  the  subjects  are  in  the  same 
community  and  hence  exposed  to  fairly  uniform  schooling. 

Lastly,  twin  resemblances  in  mental  and  in  physical 
traits  have  been,  compared.  A  similarity  in  degree  of 
correlation  in  the  two  cases  is  customarily  regarded  as 
evidence  that  the  mental  resemblances  are  independent 
of  environment.  This  is  based  on  the  argument  proposed 
by  Pearson  in  his  discussion  of  sibling  correlations.  The 
fallacy  inherent  in  such  a  conclusion  has  already  been 
discussed  in  the  preceding  chapter. 


As  early  as  1875,  Galton  (3)  compared  the  character- 
istics of  about  80  pairs  of  twins,  by  means  of  question- 
naires. The  first  quantitative  investigation  on  the  mental 


resemblance  of  twins  was  conducted  by  Thorndike  (14), 
Six  simple  mental  tests  were  administered  to  50  pairs  of 
twins  between  the  ages  of  9  and  15,  and  to  a  similar 
group  of  siblings.1  On  39  of  the  twin  pairs  physical 
measurements  were  also  obtained  for  comparative  pur- 
poses. The  correlations  on  the  entire  group  of  twins 
range  from  .69  (cancellation  of  A's)  to  .90  (naming  oppo- 
sites  of  words).  The  Correlations  are  higher,  in  general, 
in  those  tests  more  susceptible  to  home  surroundings  and 
cultural  level. 

Sibling  correlations  were  found  on  only  three  of  the  six 
tests  and  were  all  much  lower  than  the  corresponding 
twin  correlations,  their  values  being  .32,  .29,  and  .30.  The 
twins  were  further  classified  into  an  older  (age  12-14)  an<3 
a  younger  group  (age  9-11).  Th&Avgfcage  correlations^ 
.70,  for  .the  older  and  .83  for  the  younger.  Finally,  the 
average  twin  correlation  on  eight  physical  traits  proved 
to  be  .765  as  compared  to  an  average  of  .78  for  the  mental 
traits.  Thorndike  gave  a  strongly  hereditarian  interpre- 
tation to  all  of  these  findings,  relying  upon  the  arguments 
outlined  above. 

Lauterbach  (5),  who  tested  212  pairs  of  twins  with 
eight  mental  tests  and  in  addition  obtained  several  physi- 
cal measures  on  them,  reports  an  average  correlation  of 
.67  between  like-sex  2  twins  and  .41  between  unlike-sex 
or  fraternal  twins  in  mental  traits.  The  average  correla- 
tion for  older  twins  (157-328  months)  was  .55  and  for 
younger  (90-156  months)  .54.  The  resemblances  were 
higher,  on  the  whole,  in  physical  traits,  the  average  cor- 
relation in  four  such  traits  being  .77  and  .55  for  like-sex 
and  unlike-sex  groups,  respectively. 

Merriman  (8)  tested  200  pairs  of  twins  with  the  Stan- 
ford-Binet,  National  Intelligence  Test,  and  a  modified 

1  Number  and  age  not  specified.     2  Probably  identical  twins  in  most  cases. 


form  of  Army  Beta.1  Separate  correlations  were  computed 
for  like-sex  and  unlike-sex  twins,  as  well  as  for  a  younger 
(5-9  years)  and  an  older  (10-16  years)  group.  The  correla- 
tions of  the  entire  group  were  very  close  to  .80  on  all  three 
tests.  Like-sex  twins  gave  consistently  higher  correlations 
than  unlike-sex,  their  correlations  on  the  Stanford-Binet 
being  .867  and  .504,  respectively,  on  the  National  Intelli- 
gence Test  .925  and  .867,  and  on  Army  Beta  .908  and  ,732. 
In  the  majority  of  comparisons,  the  correlations  were 
higher  in  the  younger  than  in  the  older  group,  although  the 
number  of  cases  employed  in  these  correlations  was  fre- 
quently small  and  inconsistencies  were  necessarily  present. 
In  an  investigation  on  identical  and  non-identical 
twins,  Tallman  (13)  administered  the  Stanford-Binet  to 
158  pairs  of  twins  and  199  siblings  between  the  ages  of 
3  and  20.  The  average  difference  in  I.Q.  between  the 
siblings  was  13.14,  and  this  dropped  to  11.96  when  only 
siblings  who  were  less  than  two  years  apart  in  chronologi- 
cal age  were  included.  In  computing  the  average  differ- 
ence between  twins,  Tallman  combined  her  data  with 
those  of  Merriman  2  and  obtained  a  final  average  of  7.07, 
only  about  one-half  as  large  as  the  sibling  difference.  In 
this  combined  group,  there  were  84  pairs  of  unlike-sex 
twins.  The  average  I.Q.  difference  of  these  twins  was  8.48. 
The  like-sex  group  contained  178  pairs  and  gave  an  aver- 
age difference  of  642.  In  the  like-sex  group,  63  pairs  were 
classified  as  identical  twins,  on  the  basis  of  general  physi- 
cal resemblance  as  well  as  coloring  of  hair,  eyes,  and  skin. 
The  average  difference  of  these  identical  twins  was  5.08, 
as  contrasted  to  a  difference  of  7.37  for  39  non-identical 
twins  in  the  like-sex  group. 

1  Not'all  of  the  children  took  all  three  tests. 

2  Cf.  above;  105  pairs  of  twins  in  Merriman's  group  were  given  the  Stanford- 


Hirsch  (4)  reports  an  average  I.Q.  difference  of  13.8 
points  between  58  pairs  of  "dissimilar"  or  fraternal 
twins,  and  2,3  points  between  38  pairs  of  "similar"1  twins. 
Correlations  of  I.Q.'s  were  .97  for  similar  and  .53  for  dis- 
similar twins.  These  findings,  together  with  the  fact 
that  the  average  I.Q.  difference  of  4  pairs  of  similar  twins 
living  apart  was  only  3.5  points,  led  Hirsch  to  a  rather 
strongly  hereditarian  interpretation.  Only  the  results  on 
the  twins  living  apart  could,  however,  be  regarded  as 
crucial,  and  these  were  unfortunately  based  on  very  in- 
adequate data.  Apart  from  the  small  number  of  cases,  a 
serious  objection  to  these  data  is  that  the  period  of  separa- 
tion was  too  short  to  allow  much  differentiation.  The 
separation,  furthermore,  occurred  so  late  in  life  as  to  have 
practically  no  significance  in  the  development  of  the 
individual,  the  twins  in  each  pair  having  been  first  sepa- 
rated at  the  ages  of  12,  18,  29,  and  33  years,  respec- 

MeNaQHu*  (7)  administered  five  tests  of  motor  skills 
to  98  pairs  of  male  twins  in  junior  high  school.  On  the  basis 
of  physical  criteria,  46  pairs  were  clearly  classified  as 
fraternal  and  47  as  identical.  The  twin  correlations 
ranged  from  .39  to  .56  for  the  fraternal  group,  and  from 
.71  to.  .95  for  the  identical.  Continued  practice  in  three 
of  the  tests  produced  no  significant  change  in  correlation 
in  the  identical  group  and  a  rise  on  two  tests  in  the  fra- 
ternal group.  McNemar  concludes  from  these  data  that 
hereditary  factors  play  the  major  part,  in  twin  resem- 
blances in  motor  abilities.  Insofar  as  sensory  and  muscu- 
lar development  is  involved  in  such  functions,  this  finding 
is  not  especially  surprising.  In  no  case,  however,  could 
snch  a  conclusion  be  carried  over  to  otHer  kinds  of  ac- 

1  Probably  identical. 



The  study  of  twins  reared  apart  from  an  early  age 
furnishes  an  excellent  opportunity  for  the  analysis  of 
hereditary  and  environmental  factors.  If  the  twins  are 
identical,  heredity  can  be  regarded  as  constant.  The 
operation  of  similar  environmental  influences  is  limited 
to  prenatal  life  and  to  the  brief  postnatal  period  during 
which  the  twins  may  have  lived  together.  Although  not 
to  be  ignored,  the  community  of  environment  in  such 
cases  is  certainly  much  less  than  that  commonly  found 
among  twins.  Naturally,  the  number  of  twins  available 
for  observation  under  these  conditions  is  very  small.  For 
this  reason,  the  case  history  method  has  proved  the  most 
fruitful,  since  it  is  best  suited  to  a  thorough  analysis  of 
all  the  available  information  on  relatively  few  subjects. 
TJje^first^  report  of  a  pair  of  identical  twins  reared  apart 
was  published  in  1925  by  Muller  (9).  The  main  facts  of 
this  report  are  reproduced  below. 

A  pair  of  identical  female  twins  were  separated  at  the  age 
of  two  weeks  and  brought  up  by  different  families.  The  twins 
did  not  see  each  other  until  they  were  1 8  years  old,  and  even 
after  that  age  they  only  met  for  short  visits.  At  the  time  of 
examination,  they  were  30  years  of  age.  The  environment 
of  the  two  girls  presented  certain  major  differences  as  well 
as  similarities.  Twin  J  completed  high  school  and  had  some 
summer  university  work;  she  has  taught  school,  is  married, 
and  has  a  child.  Twin  B  had  only  four  years  of  formal  edu- 
cation. At  the  age  of  15,  she  took  up  clerical  work  and  has 
been  engaged  in  an  active  business  career  ever  since.  Her 
foster  family  moved  about  a  good  deal.  Twin  B  did  not 
have  as  many  social  contacts  as  her  sister.  Both  read  pro- 
fusely as  children,  and  both  had  been  energetic,  popular, 
and  active  in  club  work  in  their  communities.  On  Army  Alpha 
and  Otis  Advanced  Test  of  Intelligence,  the  twins  obtained 


very  similar  scores.  Marked  differences  were  found,  however, 
on  personality  tests,  speed  of  free  association,  and  tests  of 
motor  speed  and  coordination.  The  differences  were  in  gen- 
eral such  as  would  be  expected  from  the  variations  noted  in 
their  environments. 

Following  the  report  by  Muller,  there  have  appeared  a 
series  of  very  illuminating  case  studies  by  lawman.1  Nine 
of  these  have  been  published  to  date  and  more  are  being 
collected.  In  all  cases  the  twins  have  been  separated  from 
infancy  or  very  early  childhood.  Psychological  tests  were 
administered  in  every  instance,  so  that  some  quantitative 
data  are  available  to  supplement  the  general  observations. 
The  findings  varied  from  one  case  to  the  other.  Some 
twtfr^^wfefe^to'iincl '  who  resembled  each  other  in  most 
traits;  others  showed  close  resemblance  in  certain  char- 
acteristics and  differences  in  others ;  and  still  others  exhib- 
ite^'marked  discrepancies  in  all  traits,  intellectual,  emo- 
tional, and  even  in  such  physical  conditions  as  general 
health""  ancfbbdily  vigor. 

"Thfcse  variations  are  not  surprising.  The  accidental 
separations  of  everyday  life  to  which  a  pair  of  twins  may 
be  subjected  are  not  to  be  viewed  in  the  same  light  as  an 
experimentally  controlled  separation.  In  the  latter, 
every  effort  would  be  made  to  render  the  environments  of  the 
pair  as  unlike  as*  possible,  in  order  to  make  the  test  more 
crucial  and  the  findings  more  clear-cut.  In  the  cases  re- 
ported by  Newman,  however,  a  certain  element  of  chance 
entered  into  the  selection  of  environments.  Thus  certain 
pairs  of  twins  may  have  been  accidentally  adopted  into 
homes  which  differed  widely,  others  reared  in  surround- 
ings which  shared  a  few  important  features,  and  still 

1  For  complete  accounts,  the  reader  is  referred  to  Newman  (12).  Freeman  and 
Holzinger  cooperated  with  Newman  in  the  psychological  parts  of  these  observa- 


others  exposed  to  environments  which,  although  geo- 
graphically separate,  may  have  been  fundamentally  alike 
in  their  influence  upon  the  growing  child. 

Unless  the  cultural  milieu  and  educational  facilities  of 
the  children  are  sufficiently  different,  one  cannot  be  sure 
that  any  similarity  between  them  is  the  result  of  heredity 
alone.  This  is  particularly  true  of  a  degree  of  ability 
which  is  not  far  removed  from  the  average.  Thus,  hun- 
dreds of  children  can  be  found  with  an  LQ.  of  no,  all  of 
whom  come  from  different  families  and  live  in  different 
homes.  Nor  does  equality  of  score  on  an  intelligence  test 
indicate  mental  identity,  since  the  same  total  scote  may 
be  obtained  by  subjects  differing  in  specific  abilities.  The 
use  of  total  scores  on  complex  intelligence  tests  allows 
sufficient  leeway  to  obscure  a  moderate  amount  of  individ- 
ual variation.  The  fact  that  twins  reared  apart  from 
early  infancy  ^iay"15bTafii  very  similar  LQ-Vdpes  not, 
therefore,  constitute  proof  that  intelligence  is,  independ- 
ent of  environmental  influence.  On  the  other  hand,  it  is 
obvious  that  even  a  single  case  of  unmistakable  varia- 
tioij^between  twins  known  ,,,to  rheadentical,  in  hereditary 
constitution  is  conclusive  evidence  that  .psyqholpgical 
characteristics  are  susceptible  to  environmental  factors. 
This  is, "in 'Fact,  equivalent  to  stating  that  if  an  event 
occurs  once,  the  possibility  of  its  occurrence  has  been 

As  an  illustration  of  Newman's  findings,  a  detailed 
account  of  two  cases  is  given  below.  These  cases  were 
selected  because  the  differences  in  environment  were 
sufficiently  large  to  give  clear-cut  results. 

CASE  IV  (12,  p.  200):  Female  twins,  separated  at  the  age 
of  five  months  and  reared  by  relatives;  29  years  old  when 
examined.  Mabel  had  led  the  life  of  an  active  farm  woman 
on  a  prosperous  farm.  Mary  had  lived  largely  a  sedentary 


life  in  a  small  town,  clerking  in  a  store  during  the  day  and 
teaching  music  at  night.  Mabel  had  only  an  elementary  school 
education,  while  Mary  had  had  a  complete  high  school  course 
in  an  excellent  city  school.  At  the  time  of  examination,  a 
vast  difference  was  noted  between  the  twins  in  intellectual7 
emotional,  and  physical  traits.  Physically,  Mabel  is  described 
as  robust,  muscular,  and  in  perfect  health,  while  Mary  was 
underweight,  soft-muscled,  and  in  poor  general  condition; 
Mabel  weighed  138-5-  Ibs.,  Mary  only  nof  Ibs.  Intellectually, 
an  equally  striking  difference  was  found,  but  in  favor  of 
Mary  whose  Stanford-Binet  I.Q.  was  106  as  compared  with 
88  for  her  sister.  Even  larger  differences  were  obtained  in 
some  of  the  other  tests.  Temperamentally,  these  twins  are 
described  as  no  more  alike  than  two  persons  chosen  at 

CASE  VIII  (12,  p.  202):  Female  twins  separated  at  three 
months  of  age  and  reared  in  the  home  of  a  maternal  uncle 
and  the  latter's  brother-in-law,  respectively;  examined  when 
about  15  years  old.  M  lived  in  a  small  town  where  she  knew 
nearly  everybody  and  had  many  friends  and  playmates. 
Her  foster  father  was  well  educated  and  had  a  cultured  home, 
involving  good  books,  good  music,  etc.  R  was  brought  up 
in  a  large  city,  but  she  was  kept  closely  at  home  and  had  few 
friends.  Her  home  environment  is  described  as  narrow  and 
"unstimulating,"  neither  of  her  foster  parents  having  had 
much  education.  Formal  schooling  differed  little  in  the  two 
cases,  M  being  in  grade  loA  and  R  in  loB  at  the  time  of  exam- 
ination. The  physical  environment  is  reported  as  being  about 
the  same  for  both.  When  examined,  the  two  girls  showed  a 
remarkable  similarity  in  physical  characteristics.  Mentally, 
there  was  a  large  difference,  M  doing  consistently  better 
on  all  tests.  The  Stanford-Binet  I.Q.'s  were  92  and  77  for 
M  and  R,  respectively.  Temperamentally,  the  differences 
were  also  large,  as  shown  both  by  scores  on  personality  tests 
and  by  general  behavioral  observations.  R  is  described  as 
timid  and  retiring,  with  a  marked  lisp  in  her  speech,  and 


apparently  unhappy.     M,  on  the  other  hand,  seemed  quite 
normal  in  emotional  development. 


Another  approach  to  the  analysis  of  heredity  and  en- 
vironment has  been  made  through  the  examination  of 
large  groups  of  foster  children.  Three  such  studies  have 
been  conducted  to  date,  one  by  Burks  at  Stanford  Uni- 
versity, another  by  Freeman  and  his  associates  at  the 
Univershy  of  Chicago,  and  a  third  and  most  recent  by 
Leahy  at  the  University  of  Minnesota.  Burks  and  Leahy 
offer  rather  strongly  hereditarian  interpretations  of  their 
findings,  whereas  Freeman  lays  much  greater  stress  upon 
environmental  factors.  A  brief  analysis  of  each  study 
will  throw  some  light  upon  the  reasons  for  such  apparent 

Burks  (i)  administered  the  Stanford-Binet  to  214  foster 
children  and  382  foster  parents,  one  or  both  parents  being 
tested  in  each  family.  A  control  group  of  105  children 
and  205  natural  parents  were  similarly  examined.  The 
study  was  carefully  controlled  from  many  angles.  All  of 
the  foster  children  were  placed  in  their  adopted  homes 
under  the  age  of  12  months,  and  at  the  time  of  testing 
they  were  between  5  and  14  years  of  age.  The  control 
group  was  quite  accurately  matched  with  the  foster  group 
in  respect  to  age  of  children  and  parents,  educational, 
occupational,  and  social  status  of  the  parents,  and  cultural 
level  of  the  home.  In  both  groups,  all  subjects  selected 
for  investigation  were  White,  non-Jewish,  English-speak- 
ing Americans,  British,  or  North  Europeans,  and  all  were 
residents  of  three  districts  in  California.  Each  foster 
child  lived  in  the  home  of  a  married  couple  both  of  whom 
were  alive  and  living  together  at  the  time.  The  same 
specification  was  applied  to  the  control  group.  In  addi- 


tion  to  the  Stanford-Binet  M.A/S,  information  was  ob- 
tained on  cultural  level  of  the  home,  educational,  voca- 
tional, and  other  characteristics  of  the  parents,  home 
activities  and  care  of  the  children,  and  parents'  ratings 
on  intellectual  and  personality  traits  of  the  children. 
Data  on  the  natural  parents  of  the  adopted  children  were 
also  examined  whenever  available. 

The  correlation  between  mental  ages  of  parents  and 
children  in  both  foster  and  control  groups  is  reproduced 
below;  the  correlation  between  child's  mental  age  and 
cultural  index  of  the  home  is  also  given. 

r  between  Child's  M.A.  and:  Foster  Control 

Father's  M.A.  .07  45 

Mother's  M.A.  .19  -4^ 

Cultural  index  of  home  .25  .44 

From  these  and  similar  correlations  between  child's  M.A. 
and  various  other  conditions,  Burks  concludes  that 
heredity  is  much  more  important  than  environment  in 
determining  individual  differences  on  intelligence  tests. 
On  the  basis  of  further  statistical  analyses  of  the  data, 
she  estimates  that,  "the^total^  contribution  of  Jxeredity 
...  is  probably  not  far  from  75  or  80  per  cent"  (p.  308). 
She  also  suggests  that,  "The  maximal  contribution  of  the 
best  home  environment  to  intelligence  is  apparently  about 
20  I.Q.  points,  or  less,  and  almost  surely  lies  between  10 
and  30  points.  Conversely,  the  least  cultured,  least  stim- 
ulating kind  of  American  home  environment  may  depress 
the  I.Q.  as  much  as  20  I.Q.  points'5  (p.  309). 

. Xp.,,tjie,,  investigation  4>y-  Leaity^(6),  t]j& .same  general 
procedure  was  followed  as  in  Burks'  study,  with  a  few 
improvements.  Thus  the  Otis  Self- Administering  Test 
(Intermediate  Form)  was  substituted  for  the  Stanford- 
Binet  in  testing  the  parents,  this  test  being  better  adapted 


to  the  adult  level  than  the  Stanford-Binet.  The  match- 
ing of  the  experimental  or  adopted  group  with  the  control 
group  was  done  very  meticulously,  each  adopted  child 
being  paired  with  a  control  child  of  the  same  sex,  mental 
age  (within  6  months),  paternal  occupation,  and  father's 
and  mother's  schooling.  All  subjects  were  White,,  non- 
Jewish,  and  of  North  European  extraction.  The  children 
in  both  groups  were  reared  in  communities  of  1000  or 
more  inhabitants,  95%  of  the  group  living  in  communities 
of  over  10,000.  All  of  the  foster  children  were  legally 
adopted  by  a  married  couple.  The  age  of  adoption  was 
placed  even  lower  than  in  Burks'  study,  all  of  the  children 
in  the  experimental  group  having  been  placed  in  the  foster 
homes  at  the  age  of  6  months  or  younger.  At  the  time  of 
investigation,  the  children  ranged  in  age  from  5  to  14,  the 
average  ages  being  9.3  years  for  the  adopted  and  9.4  for 
the  control  group.  There  was  a  total  of  194  children  in 
*ach  group. 

The  general  results  of  Leahy's  study  are  in  agreement 
with  those  of  Burks.  The  correlations  of  child's  I.Q.  with 
father's  and  mother's  Otis  scores  and  with  a  cultural  index 
Df  the  home  are  reproduced  below.  Corresponding  figures 
are  shown  for  foster  and  control  groups. 

r  between  Child's  I.Q.  and:  Foster  Control 

Father's  Otis  score  .19  .51 

Mother's  Otis  score  .24  .51 

Cultural  index  of  home  .26  .51 

Various  other  comparisons  were,  made  which  lead  the 
mSTor  to  conclude,  with  Burks,,  that, .heredity ,.. js  ^the 
BatOTf  aietor  in  the  determination  of  intellectual  level.1 

J  .         ..      ,       .,., '  l  '  '  "       '          '"  ,.,          ,,        .,  „,  .       ,.u         ,     ,  .       ,--..      ,     A 

xThe  Woodworth-Mathews  Personal  Data  Sheet  for  measuring  emotional 
n'stability  was  also  administered  to  both  the  experimental  and  control  groups. 
[t  is  unfortunate,  however,  that  no  personality  measures  were  obtained  on  the 
Darents,  with  which  the  children's  scores  might  have  been  correlated.  Curiously 


In  spite  of  the  care  with  which  these  investigations  were 
conducted,  there  still  remain  too  many  disturbing  factors 
for  a  conclusive  analysis  of  hereditary  and  environmen- 
tal influences.  Thus  the  groups  tested  were  admittedly 
very  homogeneous  in  respect  to  economic,  cultural,  and 
educational  status.  Both  Burks  and  Leahy  point  out 
that  more  marked  environmental  differences  might  pro- 
duce larger  intellectual  deviations  and  that  the  results 
hold  only  for  children  brought  up  in  the  general  run  of 
American  homes. 

In  the  second  place,  the  possibility  still  remains  that  a 
group  of  children  living  with  their  natural  parents  might 
not  furnish  a  perfect  control  for  foster  children.  The  atti- 
tude of  foster  parents  towards  a  child  may  differ  in  some 
essential  ways  from  that  of  own  parents.  In  some  cases* 
the  contact  of  foster  parent  and  child  may  not  be  as  close 
or  intimate  as  that  of  a  child  and  his  natural  parents. 
The  fact  that  in  both  studies  the  foster  child's  mental 
level  correlated  higher  with  the  cultural  index  of  the  home 
than  with  mother's  intellectual  level,  and  higher  with  the 
latter  than  with  father's  intellectual  level,  whereas  all 
three  correlations  are  very  similar  or  identical  in  the  con- 
trol group  seems  to  support  such  a  suggestion.  The  child 
himself,  if  he  knows  of  his  adoption,  may  react  differently 
towards  his  foster  parents  than  he  would  towards  his  own 
parents.1  Social  expectancy  may  also  enter  into  and  corn- 
enough,  Leahy  concludes  that  environment  plays  a  far  greater  part  in  the  de- 
velopment of  personality  traits  than  in  the  case  of  intelligence.  The  basis  for 
such  a  conclusion  is  the  absence  of  significant  correlations  in  both  experimental 
and  control  groups  between  Woodworth-Mathews  score  of  child  and  parent's 
intelligence  or  cultural  index  of  home.  It  is  difficult  to  follow  the  author's  logic 
in  such  a  conclusion.  Surely  we  should  not  expect  a  very  high  correlation  be- 
tween intellectual  and  personality  traits,  regardless  of  the  hereditary  or  environ- 
mental determination  of  such  traits. 

1  In  Burks'  group,  35%  of  the  children  had  been  told  of  their  adoption,  in 
Leahy's  group,  50%, 


plicate  the  situation.  Parents  as  a  rule  expect  their  own 
children  to  resemble  them  in  intellectual  and  emotional 
development  and  this  expectation  may  be  manifested 
in  their  behavior  towards  the  child,  as  well  as  In  the  atti- 
tude of  other  relatives  and  associates.  As  the  child  de- 
velops, his  observers  constantly  call  attention  to  points 
of  family  resemblance,  real  or  imagined;  he  is  frequently 
reminded  of  ancestral  characteristics  which  are  held  up 
to  him  as  his  heritage.  Such  social  influences  are  absent 
or  greatly  reduced  in  the  case  of  foster  children.  Who 
knows  what  motivational  differences  might  thereby  arise, 
which  would  subsequently  leave  their  mark  upon  the 
development  of  abilities  ? 

In  another  extensive  investigation  on^  foster  children, 
Freeman  et  al.  (2)  present  data  from  a  variety  of  compari- 
sons. A  total  of  401  children  was  employed.  All  sub- 
jects were  residents  of  Illinois.  Both  White  and  Negro 
children  were  included,  but  the  number  of  the  latter  was 
small  The  ages  at  adoption  were  much  higher  than  in 
Burks'  study,  although  over  half  of  the  entire  group  were 
under  three  years  old  when  committed.  The  children  were 
tested  with  the  Stanford-Binet  and  the  International 
Group  Mental  Test;  1  the  foster  parents  were  given  the 
Otis  Self-Administering  Test 2  and  a  specially  constructed 
vocabulary  test  covering  many  fields  of  knowledge.  Field 
workers  collected  data  on  education,  vocation,  and  cul- 
tural level  of  the  foster  parents,  and  conditions  of  the 
foster  home.  Information  on  the  natural  parents  was 
obtained  through  visits,  interviews  with  acquaintances, 
and  examination  of  case  records. 

In  order  to  facilitate  various  comparisons,  the  children 

1  Specially  devised  as  a  universal  test  of  intelligence  and  relatively  inde- 
pendent of  specific  cultural  environment. 

2  A  common  verbal  group  test  of  intelligence. 


were  classified  into  four  overlapping  groups,  and  the  major 
results  on  each  reported  separately.  Group  I,  the  pre-test 
group.,  consisted  of  74  children  who  had  been  tested  be- 
fore adoption  and  who  had  lived  in  the  same  foster  home 
until  the  time  of  the  second  examination.  The  average 
age  of  these  children  at  adoption  was  eight  years.  The 
main  data  furnished  by  this  group  relate  to  gains  in  score 
from  the  first  to  the  second  testing.  The  average  I.Q. 
showed  a  small  but  fairly  reliable  1  rise  from  91.2  to  93.7. 
When  the  group  is  subdivided  into  those  adopted  into  the 
better  and  those  adopted  into  the  poorer  foster  homes,  the 
former  show  a  clear-cut  improvement  in  I.Q.,  the  latter 
no  change.  Similarly,  the  younger  children  proved  more 
susceptible  than  the  older  to  a  change  of  surroundings. 
The  relevant  data  are  summarized  below  (2,  p.  119). 

/•>  T       ?  D  ,-        /  v  AT  Average  I.Q. 

Cultural  Rating  of  Home  N  pim  Test  gecond  Test 

17  to  30  (Better  homes)  33  95.2  100.5 

7  to  1 6  (Poorer  homes)  41  88.0  88.1 

Age  of  Child  on  Second  Test 

12-4  or  older  37  89.7  89.3 

Under  12-4  37  92.8  98.0 

Although  the  average  gains  are  small,  the  consistency  of 

"th^gf^e^  sigmEcance.^'" '" 

Group  II,  the  sibling  group  was  composed  of  125  pairs 
of  siblings,  each  adopted  into  a  different  foster  home  and 
separated  for  a  period  of  4  to  13  years.  The  average  age 
at  which  the  siblings  became  separated  was  5  years- 

1 A  gain  of  2.5  =*=  .8,  or  over  three  times  as  large  as  its  P.E, 
2  Freeman  points  out,  furthermore,  that  the  actual  gains  are  somewhat  larger 
than  appears  from  these  figures,  since  Stanford-Binet  LQ.'s  show  a  slight  drop 
with  age  owing  to  certain  peculiarities  of  the  test  itself.  He  estimates  that  5 
points  should  be  added  to  these  gains  in  I.Q.  in  order  to  obtain  a  more  correct 


4  months.  The  I.Q.'s  of  these  siblings  correlated  only  .25, 
in  contrast  to  the  correlation  of  about  .50  ordinarily 
found  between  siblings  reared  in  the  same  home  (cf. 
Ch.  IV).  The  scores  of  63  siblings  adopted  into  homes 
receiving  significantly  different  cultural  ratings  correlated 
only  .19;  those  of  siblings  adopted  into  similar  homes 
correlated  .30.  These  results  are  particularly  striking 
when  it  is  recalled  that  the  siblings  had  lived  together 
during  the  important  years  of  early  childhood. 

The  third  group  included  all  foster  siblings,  i.e.,  two 
unrelated  children  living  in  the  same  home.  This,  In 
turn,  was  subdivided  into  a  group  of  40  pairs  consisting  of 
a  foster  child  and  an  own  child  of  the  foster  parents,  and 
a  group  of  72  pairs  of  unrelated  foster  children.  In  the 
former,  a  correlation  of  .34  was  found  between  I.Q.'s  of 
the  two  children  in^each  pair;  in  the  latter,  the  correlation 
was  .37.  These  correlations  are  actually  higher  than  those 
between  true  siblings  adopted  into  different  foster  homes. 

Finally,  all  of  the  children  were  included  in  one  com- 
posite group  of  401  cases.  This  composite,  labeled  the 
home  group  by  Freeman,  was  employed  chiefly  in  making 
general  comparisons  between  foster  child's  intelligence 
and  social  adjustment,  and  such  factors  as  foster  parents' 
intelligence  and  cultural  level  of  the  foster  home.  ^In 
th£^H,tke-. group,  a  correlation,  ol  .48^  was  ..found , between  ' 
child's  I.Q.  and  cultural  rating  of  the  foster^home.  The 
correlation  of  child's  I.Q.  with  Otis  score  of  the  foster 
father  was  .37  (N  =  1 80)  and  with  that  of  the  foster 
mother  .28  (N  —  255).  These  and  other  similar  correla- 
tions again  suggested  the  relatively  large  influence  of 
environment  upon  intelligence  test  performance.1 

1  The  specific  results  cited  in  the  above  discussion  are  all  based  on  the  Stan- 
ford-Binet  and  the  Otis  tests.  The  International  and  the  vocabulary  tests 
yielded  very  similar  results  in  all  cases  in  which  they  were  employed. 


The  analyses  carried  out  in  the  four  groups  yield  results 
which  corroborate  each  other  closely.  The  data  appear 
very  conclusive.  A  difficulty  in  the  way  of  a  clear-cut 
interpretation  is,  however,  the  presence  of  a  possible 
selective  factor  in  adoption.  Thus  the  more  intelligent 
persons,  with  more  cultured  homes,  may  tend  to  adopt 
the  more  intelligent  children.  This  would  produce  an 
entirely  spurious  correlation  between  intelligence  of  foster 
parents  and  children  as  well  as  between  cultural  level  of 
the  home  and  child's  I.Q.  Freeman  recognizes  this  diffi- 
culty, but  attempts  to  show  that  such  selection  is  very 
insignificant.  An  examination  of  the  application  blanks 
submitted  by  the  foster  parents  showed  that  factors 
other  than  intelligence  were  usually  considered.  Health, 
sex,  race,  and  physical  appearance  were  more  often  speci- 
fied than  intellectual  level.  When  the  latter  was  men- 
tioned, it  was  only  to  require  that  the  child  be  of  "normal" 
intelligence,  but  no  more  specific  account  was  taken  of 
degree  of  intelligence.  This  request,  furthermore,  was 
made  equally  often  by  people  who  were  themselves  in- 
tellectually inferior  as  by  those  who  were  superior.  Fi- 
nally, in  over  80%  of  the  cases,  very  little  information 
was  available  at  the  time  of  adoption  about  the  intelli- 
gence of  the  child  or  of  his  natural  parents.  In  the  case 
of  the  74  children  for  whom  I.Q.'s  were  available  before 
adoption  (cf.  above),  there  is  evidence  that  such  I.Q.'s 
were  employed  in  placing  the  children,  "but  this  group 
constituted  only  a  small  portion  of  the  total  group  0/401  cases. 

If  the  studies  of  Burks,  Leahy,  and  Freeman  are  evalu- 
ated together,  it  appears  that  the  role  of  environment  in 
the  development  of  intelligence  cannot  be  dismissed. 
Beyond  this,  however,  it  is  impossible  to  generalize  be- 
cause of  special  difficulties  inherent  in  each  study.  In 
Freeman's  investigation,  the  factor  of  selection  may  have 


operated,  although  Freeman  asserts  that  Its  effect  was 
probably  negligible.  In  Burks'  and  Leahy's  studies,  in 
which  the  age  of  adoption  was  much  lower  and  selective 
factors  were  thus  more  completely  eliminated,  the  only 
comparisons  made  were  not  very  conclusive.  As  has 
already  been  pointed  out,  the  evaluation  of  the  foster 
data  in  terms  of  the  control  data  was  not  perfectly  justifi- 
able, since  the  parent-child  relationships  were  probably 
dissimilar  in  the  two  groups.  It  is  unfortunate  that  Burks' 
and  Leahy's  groups  of  foster  children,  with  their  obvious 
advantage  of  lower  adoption  age,  could  not  have  been 
submitted  to  the  same  varied  analyses  employed  by 
Freeman.  A  comparison  of  siblings  adopted  under  the 
age  of  one  year  into  different  foster  homes,  for  example, 
would  have  been  very  valuable.  It  would  have  been 
quite  difficult,  however,  to  find  a  sufficiently  large  number 
of  such  siblings  meeting  the  required  specification  in 
respect  to  adoption  age.  If  one  requirement  is  met, 
another  must  be  sacrificed.  It  would  seem,  then,  that  the 
study  of  foster  children  has  not  yet  proved  as  fruitful  an 
approach  to  the  problem  of  heredity  and  environment  as 
it  had  promised  to  be. 


1.  Burks,  B.  S.    "The  Relative  Influence  of  Nature  and  Nur- 
ture upon  Mental  Development;  a  Comparative  Study  of 
Foster  Parent-Foster  Child  Resemblance  and  True  Parent- 
True  Child  Resemblance/5  2jth  Yearbook,  Nat.  Soc.  Stud, 
Educ.,  1928,  Part  I,  219-316. 

2.  Freeman,  F.  N.,  Holzinger,  K.  J.,  and  Mitchell,  B.  C.  "The 
Influence    of    Environment    on    the    Intelligence,    School 
Achievement,  and  Conduct  of  Foster  Children,"  2?th  Year- 
book, Nat.  Soc.  Stud.  Educ.,  1928,  Part  I,  103-217. 

3.  Galton,  F.    Inquiries  into  Human  Faculty  and  Its  Develop- 
ment.  London:  Macmillan,  1883.  Pp.  387. 


4.  Hirsch,  N.  D.  M.    Twins:  Heredity  and  Environment.    Cam- 
bridge: Harvard  Univ.  Press,  1930.    Pp.  158. 

5.  Lauterbach,  C.  E.     "Studies  in  Twin  Resemblance,"  Ge- 
netics ,  1925,  10,  525-568. 

6.  Leahy,  A.  M.    "Nature-Nurture  and  Intelligence,"  Genet. 
Psychol.  Mon.,  1935,  17,  236-308. 

7.  McNemar,  Q.     "Twin  Resemblances  in  Motor  Skills,  and 
the  Effect  of  Practice  Thereon/'  /.  Genet.  Psychol.,  1933, 
42,  70-99. 

8.  Merriman,  C.    "The  Intellectual  Resemblance  of  Twins," 
Psychol.  Hon.,  1924,  33,  No.  5.   Pp.  58. 

9.  Muller,  H.  J.     "Mental  Traits  and  Heredity,"  /.  Hered., 

10.  Newman,  H.  H.     The  Biology  of  Twins.     Chicago:  Univ. 

Chicago  Press,  1917.  Pp.  185. 
IIt  -  .    The  Physiology  of  Twinning.   Chicago:  Univ.  Chicago 

Press,  1923.  Pp.  230. 

12.  -  .    "The  Effects  of  Hereditary  and  Environmental  Dif- 
ferences upon  Human  Personality  as  Revealed  by  Studies 
of  Twins,"  Amer.  Naturalist,  1933,  67,  193-205. 

13.  Tallman,  G.  G.    "A  Comparative  Study  of  Identical  and 
Non-Identical  Twins  with  Respect  to  Intelligence  Resem- 
blances," 27th  Yearbook,  Nat.  Soc.  Stud.  Educ.,  1928,  Part  I, 

£4.  Thorndike,  E.  L.  "Measurement  of  Twins,"  Arch.  Psychol., 
1905,1.  Pp.6^ 


A  distinction  has  frequently  been  made  between  de- 
velopment through  specific  practice  or  training  in  a  given 
activity  and  development  through  maturation  or  growth. 
Such  a  distinction  does  not  imply  a  dichotomy  between 
inherited  and  acquired  behavior.  Thus  maturation  is  not 
regarded  as  independent  of  environmental  stimulation  of 
a  general  sort,  nor  is  learning  necessarily  considered  to  be 
exclusively  determined  by  environmental  factors.  When 
we  speak  of  growth,  we  usually  think  of  a  definite  se- 
quence of  developmental  stages  in  the  structural  char- 
acteristics of  the  individual.  As  the  child  grows  older, 
for  example,  his  height  increases,  his  bodily  proportions 
are  altered,  and  many  other  well-known  physical  modifi- 
cations occur.  Such  changes  take  place  regardless  of  the 
specific  training  which  the  individual  may  have  had. 

As  structures  become  altered  with  age,  so  we  may  expect 
their  functions  to  undergo  change.  With  stronger  muscles, 
the  older  child  can  learn  to  walk,  climb  stairs,  sit  up,  and 
perform  various  other  tasks  much  more  readily  than  his 
younger  brother.  It  is  logical  to  expect  that  certain 
types  of  activity  will  in  general  appear  at  fairly  definite 
stages,  since  they  require  a  specific  degree  of  structural 
development  for  their  execution.  Very  intensive  training 
at  an  earlier  age  may  produce  almost  negligible  effects 
when  compared  to  the  achievements  of  an  older  child 
with  only  a  minimum  of  training. 

Since  such  a  large  share  of  infant  behavior  consists  in  the 
acquisition  of  motor  skills  and  sensori-motor  coordinations, 



activities  which  are  closely  linked  to  structural  factors, 
growth  rather  than  practice  seems  to  play  the  major  part 
in  early  behavioral  development.  It  is  quite  a  different 
matter,  however,  to  use  the  concept  of  growth  to  describe 
the  intellectual  and  emotional  development  of  the  older 
child.  Such  a  concept  has  nevertheless  been  commonly 
employed  in  interpreting  age  changes  in  mental  test  per- 
formance, and  the  curves  plotted  to  portray  these  changes 
graphically  have  been  labeled  " mental  growth  curves." 
Such  growth  curves  are  difficult  to  interpret  for  many 
reasons  and  their  use  has  led  to  much  technical  controversy. 

An  equally  confused  issue  centers  about  the  question 
of  the  effect  of  growth  or  practice  upon  the  extent  of 
individual  differences.  Do  people  differ  more  among 
themselves  ,when  their  general  level  of  performance  is 
high  of  when  it  is  low?  Will  subjects  become  more  alike 
or  more  unlike  with  training?  Is  variability  greater  among 
older  or  among  younger  individuals?  In  all  of  these 
phases  of  the  problem  of  training  and  growth  in  relation 
to  individual  differences,  there  have  appeared  numerous 
investigations.  But  the  results  are  very  inconclusive  and 
frequently  misleading. 

In  the  present  chapter  will  be  discussed  those  investiga- 
tions which  are  primarily  concerned  with  the  effects  of 
specific  training  upon  performance.  In  such  studies,  some 
effort  is  made  to  bring  about  a  change  in  behavior  under 
experimentally  controlled  conditions.  Those  investigations 
which  simply  present  observations  or  measurements  made 
on  different  age  groups  without  any  attempt  to  alter  the 
course  of  development  will  be  given  in  the  following  chapter. 


In  the  course  of  extensive  and  carefully  controlled  ob- 
servations and  experiments  conducted  at  the  Yale  Psycho- 


Clinic,  Gesell  *  has  established  developmental  norms  of 
infant  behavior  for  different  ages.  Since  the  behavior 
repertory  of  the  child  during  the  first  few  years  of  life 
consists  so  largely  in  the  exercise  of  simple  sensory  and 
motor  functions,  most  of  the  data  are  drawn  from  this 
type  of  activity.  Gesell  concludes  from  various  findings 
that  such  activities  depend  chiefly  upon  growth  or  matura- 
tional  factors.  Among  his  evidence  he  cites  observations 
on  pre-term  and  post-term  babies.  Infants  born  pre- 
maturely, before  the  normal  nine-month  gestation  period, 
do  not  reach  the  developmental  level  of  the  normal  new- 
born child  until  the  age  of  one  month.  Similarly,  a  baby 
born  after  a  ten-month  gestation  period  will  be  as  far 
advanced  in  its  behavior  at  birth  as  a  normal  one-month 
old  child.  Yet  the  differences  in  opportunity  for  specific 
practice  in  prenatal  and  postnatal  environment  are  ob- 
viously large. 

Qesell  likewise  points  to  the  consistency  of  develop- 
mental sequences,  as  evidence  for  maturation.  In  the 
development  of  prehension  behavior,  for  example,  the 
successive  stages  come  in  the  same  order  and  at  approx- 
imately the  same  ages  in  different  children.  Observation 
of  the  child's  reactions  towards  a  small  sugar  pellet  showed 
that  both  in  visual  fixation  and  in  hand  and  finger  move- 
ments, characteristic  behavior  was  displayed  at  successive 

Training  experiments  also  tended  to  corroborate  the 
same  general  view.  A  series  of  experiments  were  con- 
ducted by  the  "method  of  co-twin  cpntrpl"  whereby  one 
member  of  a  pair  of  identical  twins  is  subjected  to  inten- 
sive training  in  some  activity  while  the  other  is  used  as  a 
control  subject  and  prevented  from  exercising  the  func- 
tion under  investigation.  In  one  such  experiment  (7, 

1  Cf.,  e.g.,  6  and  7. 


p.  654),  stair-climbing  and  "cube  behavior"  (including 
prehension,  manipulation,  and  constructive  play  with 
cubes)  were  studied  on  a  pair  of  identical  female  twins, 
46  weeks  old  at  the  beginning  of  the  experiment.  The 
trained  twin  (T)  was  put  through  a  daily  2ominute 
training  period  in  both  types  of  activity  for  six  weeks. 
At  the  end  of  this  period,  the  control  twin  (C)  who  had 
had  no  specific  training  in  these  functions  proved  equal 
to  T  in  cube  behavior.  In  stair-climbing,  a  difference 
was  found.  Whereas  T  was  a  relatively  expert  climber, 
her  sister  could  not  reach  the  top  of  a  five-tread  staircase 
even  ^ith  assistance.  Two  weeks  later,  however,  still 
without  any  previous  training,  the  control  twin  was  able 
to  climb  to  the  top  unassisted.  At  this  age  (53  weeks), 
twin  C  was  herself  given  a  two-week  training  period,  at 
the  end  of  which  she  approximated  T  in  her  climbing 
skill.  Thus,  because  of  the  higher  level  of  maturational 
development,  a  two-week  training  period  at  53  weeks  of 
age  proved  to  be  nearly  as  effective  as  a  six-week  training 
period  at  46  weeks. 

Experiments  on  intensive  training  in  infancy  have  also 
been  conducted  by  McGraw  (10).  A  pair  of  male  twins  1 
were  observed  from  birth  to  the  age  of  22  months.  Jimmy, 
the  twin  who  appeared  stronger  and  better  developed  at 
birth,  served  as  the  control,  his  activity  being  approxi- 
mately that  of  a  normal  infant  during  the  earlier  period 
and  possibly  a  little  more  restricted  than  normal  later. 
The  other  twin,  Johnny,  was  put  through  intensive  daily 
training  from  the  age  of  20  days.  Both  twins  lived  at 
home  but  were  in  the  laboratory  between  9  and  5  o'clock 
for  five  days  a  week.  The  performance  of  the  trained 
twin  in  each  task  was  compared  throughout  the  period 

1  Originally  believed  to  be  identical  although  subsequent  physical  character- 
istics threw  doubt  upon  this  designation. 


of  the  experiment  with  that  of  the  untrained  control  twin. 

Specific  exercise  was  found  to  have  little  or  no  effect 
upon  a  group  of  activities  including  simple  reflexes,  such 
as  suspension-grasping,  as  well  as  crawling  and  creeping, 
erect  walking,  sitting,  prehension,  and  others.  Marked 
improvement  resulted,  however,  from  practice  on  a  group 
of  somewhat  more  complex  functions  such  as  skating, 
jumping,  swimming,  diving,  ascending  and  descending 
inclines,  getting  off  stools,  and  manipulating  and  climbing 
stools  and  boxes  to  reach  an  objective.  Although  a  certain 
amount  of  sensory  and  muscular  development  obviously 
helps  in  the  latter  functions,  their  performance  seems  to 
depend  largely  on  specific  training.  The  independence 
of  the  former  group  of  functions  from  practice  confirms 
many  of  GeselFs  findings.  For  the  execution  of  these 
simpler  functions,  the  presence  of  structures  of  a  certain 
degree  of  development  seems  to  be  sufficient  or  nearly 

In  an  experiment  by  Dennis  (4),  the  opposite  procedure 
of  restricting  training  was  followed.  Two  female  infants, 
who  happened  to'be  fraternal  twins  although  this  was  not 
essential  to  the  present  experiment,  were  reared  in  the 
experimenter's  home  under  controlled  conditions  from 
one  to  14  months  of  age.  During  this  period,  all  oppor- 
tunity to  stand  or  sit  was  "eliminated  and  opportunities 
to  grasp  objects  were  reduced  to  a  minimum.  Comparing 
the  behavior  of  these  infants  at  successive  periods  of  the 
experiment  with  norms  established  by  Gesell  and  others, 
Dennis  found  marked  retardation  in  these  functions.  The 
infants  W£re  unable  to  stand,  sit  unsupported,  or  grasp 
a  dangling  ring  at  the  ages  reported  as  "normal"  for  each 

These  experiments  indicate  some  of  the  major  trends 
revealed  by  the  study  of  infant  behavior.  There  are  some 


activities  which  depend  so  largely  upon  structural  condi- 
tions, such  as  sensory,  muscular,  or  neural  development, 
that  their  performance  is  almost  completely  unaffected 
by  specific  exercise.  When  more  complex  and  less  struc- 
turally fixed  activities  are  investigated,  however,  the  effect 
olLtraining  is  apparent.  It  should  also  be  borne  in  mind 
in  interpreting  the  above  experiments  that  conditions 
were  never  perfectly  controlled.  For  practical  reasons,  it 
is  impossible  either  to  determine  or  to  observe  all  the 
stimulation  to  which  the  infant  is  exposed.  In  McGraw's 
study,  for  example,  no  control  was  exerted  over  the  in- 
fants' activities  during  the  time  they  spent  at  home. 
Finally,  even  if  specific  practice  is  rigidly  prevented,  the 
influence  of  the  more  or  less  random  and  general  training 
which  the  infant  obtains  in  the  course  of  everyday  activity 
cannot  be  ignored.  Thus,  for  example,  the  greater  effec- 
tiveness of  training  in  stair-climbing  when  applied  at  a 
later  age  may  result  as  much  from  the  fact  that  the  older 
child  has  used  his  muscles  more  in  various  general  activ- 
ities, as  from  his  more  advanced  age. 

Mention  should  also  be  made  of  experiments  on  infra- 
human  organisms,  both  by  the  method  of  special  training 
and  that  of  restricted  practice.  In  their  general  implica- 
tions, the  findings  are  very  similar  to  those  on  human 
infants.  Certain  functions,  such  as  the  flying  of  birds  (13) 
or  the  swimming  movements  of  tadpoles  (2),  for  example, 
have  been  found  to  depend  almost  exclusively  upon  struc- 
tural development.  Other  activities  require  a  moderate 
amount  of  specific  exercise  together  with  a  certain  level 
of  physical  development  for  their  normal  execution;  still 
others  depend  very  largely  upon  the  nature  of  the  train- 
ing and  specific  stimulation  to  which  the  organism  is 
subjected,  and  are  affected  only  in  a  very  general  way 
by  maturational  factors. 



Attempts  have  also  been  made  to  investigate  the  effect 
of  specific  training  upon  functions,  such  as  memory  or 
intelligence  test  performance,  which  are  more  complex 
and  less  directly  related  to  sensori-motor  development. 
Most  of  these  experiments  have  been  conducted  on  chil- 
dren of  elementary  school  age.  Gates  (5)  studied  the 
effect  of  continued  practice  upon  memory  span  for  digit§. 
Two  groups  of  schoolchildren,  selected  so  as  to  be  equiv- 
alent in  age,  number  of  boys  and  girls,  Stanford-Binet  I.Q., 
school  grade,  teachers'  estimates  of  scholastic  maturity, 
and  scores  on  several  memory  tests,  were  given  an  initial 
test  in  digit  span.  The  children  in  the  Practice  group 
were  then  put  through  individual  practice  in  recalling 
digits  on  each  of  78  days  extending  over  a  period  of  five 
months.  At  the  end  of  this  period,  both  Practice  and 
Control  groups  were  given  a  final  test.  The  average 
scores  on  initial  and  final  tests  are  reproduced  in  Table  IV. 


(After  Gates,  5,  pp.  454~456) 


Initial  Test 

Final  Test 

After  a  Lapse 
of  4%  Months 

After  Common 
Practice  for 
22  Days 







^  Jbut^the  ^Practice  group 

is  clearly  ahead,  manifesting  a    ^ia.whick.aozmji       re- 

a  six-year  period,  according  to  the  Stanford-Binet 
norms  for  this  function.  Four  and  one-half  months  after 
the  SnaTtes^5oSL  groups  were  again  tested,  by  a  differ- 


ent  examiner.  This  time  the  Practice  and  Control  groups 
were  approximately  equal  Finally,  the  two  groups  were 
subjected  to  22  days  of  practice,  at  the  end  of  which  both 
showed  improvement  and  in  approximately  equal  degree. 
It  was  also  found  that  the  training  in  digit  span  had  no 
effect  on  performance  in  other  types  of  mental  tests. 

From  these  findings,  Gates  concludes  that  training  is 
highly  specific,  consisting  in  the  acquisition  of  special 
skills  and  techniques,  and  that  it  does  not  alter  the 
growth  of  the  underlying  mental  functions.  It  is  unfor- 
tunate that  in  most  studies  on  experimentally  adminis- 
tered practice,  all  effects  not  directly  resulting  from  such 
practice  are  attributed  indiscriminately  to  growth  or  mat- 
urational  phenomena,  the  influence  of  the  vast  amount 
of  other  training  which  the  child  is  receiving  in  the  course 
of  everyday  life  being  disregarded. 

In  Gates'  experiment,  the  fact  that  the  Control  group 
also  improved,  although  to  a  lesser  extent  than  the  Prac- 
tice group,  suggests  the  effect  of  other  intervening  experi- 
ences rather  than  growth.  This  explanation  is  further 
corroborated  by  the  finding  that  both  groups  drop  to  an 
equal  level  when  retested  later  by  another  examiner.  The 
drop  may  have  resulted  from  the  time  of  year  at  which 
the  tests  were  administered,  or  from  other  factors  inci- 
dental to  the  school  situation.  The  closeness  with  which 
the  child  attends  to  the  material  and  the  effort  he  puts 
forth  to  concentrate  on  the  task  of  memorizing  are  very 
important  factors  in  determining  his  span;  and  it  seems 
entirely  plausible  that  such  factors  should  be  affected 
both  by  the  attitude  of  the  particular  examiners  and  by 
the  sum  total  of  school  experiences  which  the  child  has 
had.  It  is  noteworthy  that  the  4|-month  period  preceding 
the  drop  in  score  included  the  summer  vacation,  which  is 
definitely  an  environmental  and  not  a  maturational  inci- 


dent.  The  marked  susceptibility  of  a, function  like  memory 
span  to.  training,  which  this  experiment  demonstrated, 
seems  in  itself  to  minimize  maturational  factors.  To  as- 
sume the  existence  of  some  underlying  hypothetical  capac- 
ity of  memorizing  which  remains  unaltered  while  per- 
formance on  a  memory  span  test  rises  and  falls  seems 
totally  unwarranted  by  the  facts  and  not  a  very  clarifying 
procedure  in  any  event. 

Several  investigations  have  been  conducted  to  deter- 
mine the  effect  of  special  training  or  coaching  upon 
Stanford-Binet  scores.  In  three  studies  (3)  carried  out 
under  the  direction  of  Terman  at  Stanford  University, 
children  were  given  instruction  and  practice  for  several 
weeks  on  material  either  identical  or  similar  to  some  of 
the  tests  in  the  Stanford-Binet  scale.  The  groups  were 
small,  varying  from  10  to  26,  but  in  each  study  the  trained 
group  was  carefully  matched  with  a  control  group  by 
"pairing"  the  subjects.  All  experiments  clearly  demon- 
strated the  possibility  of  teaching  a  child  to  perform 
tests  which  he  was  formerly  incapable  of  doing  because  of 
age  or  mental  level.  The  influence  of  this  improvement 
upon  the  LQ.  obtained  on  the  whole  scale  differed  in  the^ 
three  studies,  being  most  evident,  as  would  be  expected, 
in  that  study  l  in  which  the  trained"  functions  overlapped 
with  the  largest  number  of  Stanford-Binet  tests.  In  this 
st^dy,  furthermore,  retests  after  a  six-week  period,  during^ 
whicH  lieitKef  group  had  received  any  trajjung,  jshpwfc&the 
practice  group  to  have  retained  its  advantage  over  the 
control  group. 

WYCS.tigatiQix  on  the  .effects,,  of,  coaching 
is  reported  by  Greene  (8).  Three  groups  of  children  were 
given  the  Stanford-Binet.  The  subjects  in  one  of  these 
groups  were  then  coached  on  the  specific  tests  in  which 

1  I.e.,  the  study  by  Casey  (3,  pp.  431-433). 


they  had  failed.  A  second  group  was  coached  on  material 
similar  but  not  identical  to  that  in  the  Stanford-Binet. 
No  child  in  either  group  was  coached  for  over  two  hours 
altogether.  The  third  group  served  as  a  control,  receiving 
no  special  training  in  the  test  material.  All  groups  were 
retested  at  intervals  of  three  weeks,  three  months,  one 
year,  and  three  years  after  the  initial  tests.  The  average 
I.Q.'s  of  each  group  on  the  initial  test  and  each  of  the 
four  retests  are  given  below  (8,  p.  425).  The  results 
obtained  in  two  schools,  A  and  Y,  have  been  kept  sep- 
arate since  a  slightly  different  method  of  coaching  was 
employed  in  each. 

School  A  School  Y 

Test  Control  Coached     Similar          Control     Coached     Similar 

(N  =  9)  (N  =  1 8)  (N  =  17)      (N  =  17)  (N  =  11)  (N  »  16) 











3  weeks 









3  mos. 









I  year 









3  years 





1  02 



It  will  be  noted  that  whereas  the  control  groups  show 
only  irregular  fluctuations  from  time  to  time,  the  coached 
groups  in  both  schools  undergo  a  marked  improvement 
oji  the  second  test  which  followed  shortly  after  the  coach- 
ing period.  This  improvement  is  retained  on  successive 
retests,  although  in  ever  decreasing  amount.  The  grad- 
ual drop  in  I.Q.  observed  in  the  coached  groups  may  be 
owing  partly  to  forgetting  of  the  coached  material  and 
partly  to  the  fact  that,  as  the  children  grew  older,  they 
were  tested  to  an  increasing  extent  at  higher  age  levels 
in  which  they  had  not  been  coached.  That  the  latter  is  prob- 
ably the  major  factor  is  demonstrated  by  a  comparison 
of  the  coached  groups  in  the  two  schools.  In  school  A, 
the  children  were  coached  more  intensively  on  fewer 
tests;  in  school  Y,  they  were  coached  on  two  additional 


higher  levels.  Thus  the  effects  of  coaching  in  school  Y 
should  not  be  "outgrown"  as  readily  as  in  A.  The  aver- 
age I.Q.'s  do  in  fact  show  that  the  dropping  off  does  not 
occur  as  rapidly  in  Y  as  in  A.  The  groups  trained  on 
similar  material  also  show  an  immediate  improvement 
which  gradually  disappears  on  successive  retests.  As 
would  be  expected,  the  gains  in  these  groups  are  much 
smaller  throughout  than  in  the  groups  which  had  been 
directly  coached. 

All  of  these  studies  indicate  the  very  great  extent  to 
wElcE""l]aeatal  test  performance  may  be  influenced  by 
training.  Such  findings  suggest  vast  possibilities  regarding 
the  part  played  by  the  incidental  and  often  accidental 
training  of  everyday  life.  That,  the  effects,  of  a  brief 
period  of  training  are  not  permanent  seems  to  be  quite 
beside  the  point.  When  training  is  discontinued,  we  should 
naturally  expect  the  improvement  to  fall  off  because  of 
forgetting.  If,  furthermore,  children  are  tested  in  different 
functions  at  successive  ages,  as  they  are  to  a  large  extent 
in  the  Stanford-Binet,  the  effects  of  training  will  not  be 
manifested  over  a  long  period.  It  is  futile  to  expect  that 
a  brief  period  of  highly  specific  instruction  or  practice 
should  raise  the  "general  mental  level"  of  the  child, 
especially  since  such  a  mental  level  is  itself  a  manifold 
of  widely  diverse  and  loosely  interrelated  functions.  Train- 
ing does  have  a  very  real  effect,  however,  upon  the  indi- 
vidual's performance  on  specific  mental  tests.  And  this 
is  of  prime  importance  since  all  our  observations  regarding 
the  subject's  psychological  make-up  are  ultimately  derived 
from  such  concrete  behavior. 


about  a  pronounced  change  in  giej^l  test,  performance,  a 


further  question  may  be  raised  regarding  the  differential 
effects  of  such  training  upon  individual  subjects.  Will 
the^  initially  better  individuals  benefit  more  than  Tfie 
initially  poorer?  Will  subjects  tend  to  maintain  the  same 
relative  standing  in  the  course  of  training?  .E&Jndjyidual 
differences  increase  or  decrease  with  practice?  If  these 
questions  are  still  unanswered,  it  is  not  for  dearth  of  data, 
for  they  have  been  repeatedly  investigated  with  a  wide 
variety  of  materials,  methods,  and  subjects.1  The  entire 
problem  is  so  beset  with  technical  difficulties,  however, 
as  to  have  even  been  declared  insoluble  by  some.  The 
crux  of  the  matter  is  that  entirely  opposite  conclusions 
can  be  drawn  if  the  results  are  expressed  in  different 
forms,  a  fact  which  has  cast  an  aura  of  artificiality  over 
all  the  data. 

In  the  present  section  will  be  given  a  brief  survey  of  the 
major  issues  involved  in  the  problem  of  practice  and  vari- 
ability. These  must  be  considered  before  any  attempts 
are  made  to  examine  the  particular  findings.  The  data 
are  meaningless  unless  evaluated  in  terms  of  the  specific 
questions  which  we  wish  to  answer  and  the  methodology 
necessitated  by  such  questions.  This  section  may  seem 
somewhat  of  a  technical  digression,  but  it  cannot  be 
eliminated  from  any  analysis  of  the  effects  of  practice 
upon  individual  differences.  Attempts  to  present  only  a- 
simplified  summary  of  results  have  proved  exceedingly 
misleading,  since  the  reviewer  in  such  cases  must  either 
arbitrarily  omit  many  of  the  data  or  offer  conflicting 
conclusions  with  no  possibility  of  reconciling  them. 

, 2! A£ J^lffipwlti^  Q£  -this,  probkni,  are,,  inherent  in 
arison  of  yan.ability^jsither ^ from,, trait , tpjjait 
(cf .  Ch.  II)  ^or  from  one  condition  to  another^  As  is  true 

xFor  summaries  of  the  relevant  literature,  the  reader  is  referred  to  Kin- 
caid  (9),  Peterson  and  Barlow  (n),  Reed  (12),  and  Anastasi  (i). 


in  all  of  these  cases,  if  a  solution  is  to  be  found  it  must 
be  stated  in  terms  of  a  specifically  defined  situation. 
Much  of  the  controversy  and  confusion  seems  to  have 
arisen  from  the  attempt  to  go  beyond  the  concretely 
established  facts  and  discuss  a  sort  of  disembodied  ab- 
stract "  variability  "  which  is  expected  to  be  independent 
of  the  particular  situation  in  which  it  has  been  measured. 
In  any  analysis  of  the  effect  of  practice  upon  individual 
differences,  it  is  necessary  to  ascertain  at  the  outset  wjiat 
is  meant  by  equal  practice.  If  all  in^mduals,^re,peimit,ted 
to  practice  for  ^13.  equal  period  of  time,  the  slower  .worker 
will  be  at  a  disadvantage  since  he  will  have  received 
practice  on  less  material  than  the  faster  individual.  The 

**.."..,..,,..  ;  ,     '  ,  ",  ***»•», 

use  of  an  equal  amount  ,pf ,  rriaterial,  on  the  other  hand, 
places  a  handicap  on  the  faster  worker  who  will  neces- 
sarily spend  less  time  in  training  than  the  slower  perspn. 
Tljie  amount  /zwi^methodj^iying  the^  advantage  to  .vthe^ 
initially  poorer  individual,  fayors  a  decrease  in  .variability 
with  practice,  whereas  the  time  limit  method  favors,,,  an 

Each  method  answers  a  somewhat  different  question. 
The  best  criterion  for  choosing  between  the  two  seems 
to  be  a  practical  one.  Equal  training,  as  the  term  is 
used  in  everyday  life,  usually  refers  to  equal  time  spent 
in  training.  When  a  person  takes  a  "course"  in  music, 
or  golf,  or  Spanish  conversation,  he  is  given  a  specified 
number  of  lessons,  each  of  the  same  duration.  No  adjust- 
ment is  made  for  the  fact  that  during  that  period  the 
number  of  times  a  piano  key  is  touched  or  a  golf  ball  is 
hit,  or  the  number  of  words  spoken  differs  widely  from 
one  individual  to  another. 

I*1  Table  V  are 
illustrated  three  alternative  ways  of  reporting  the  same 



I.  Amount  Scores 



Number  of  Items  Completed 
during  a  i-Minute  Trial 

Gain  in  Items 
-per  Trial 



First  trial 

Last  trial 





2.   Time  Scores 



Average  Time  in  Seconds 
to  Complete  One  Item 

Gain  in  Time 
per  Item 



First  trial 

Last  trial 



3  .   Time  Saved  p  er 



Gain  in  Items 
per  Trial 


Time  Initially 
Each  Item 

Gain  in  Time 
per  Trial 


scores  obtained  by  two  subjects,  A  and  B,  with  the  time 
limit  method.  In  this  table,  A  represents  an  initially  faster 
worker  and  B  an  initially  slower  one.  It  will  be  noted 
that  the  relation  between  the  gains  of  the  initially  better 
and  poorer  subjects  differs  in  each  method.  When  the 
scores  are  expressed  as  amount  of  work  done  per  unit  of 
time,  the  gain  of  the  better  subject  appears  larger  than 
that  of  the  poorer  one.  This  will  tend  to  make  variability 
increase  with  practice.  If,  on  the  other  hand,  these  same 
scores  are  expressed  as  time  per  unit  of  work,  the  slower 
individual  will  seem  to  gain  more. 

This  apparent  contradiction  becomes  intelligible  if  we 
realize  just  what  time  and  amount  scores  are^nxqasuring. 


for  ^  every   additional  item  which  he  completes  within^  the 


given  period  after  practice  he  will  be  saving  much  more  time, 
than^  the  faster  worker.  Thus,  if  It  took  the  slow  worker  B 
5  seconds  to  complete  one  item  at  the  beginning  of  prac- 
tice and  if  he  can  complete  8  more  items  after  practice 
than  he  could  before,  he  has  gained  the  equivalent  of 
8  X  5  or  40  seconds  per  trial  The  faster  worker  A3  on 
the  other  hand,  added  10  items  to  his  score,  but  he  only 
required  3  seconds  per  item  at  the  outset,  so  he  has  gained 
10  X  3  or  30  seconds  (cf.  method  3,  Table  V).  The  gain 
in  time  per  item  (method  2,  Table  V)  favors  the  slower 
worker  even  further,  since  it  does  not  take  into  account 
the  fact  that  during  any  one  trial  this  unit  gain  in  speed 
is  manifested  more  often  by  the  faster  than  by  the  slower 
worker,  the  former  completing  more  items. 

It  is  apparent,  then,  that  the  problem  of  practice  and 
variability  must  be  further  defined  in  terms  of  the  measujre 
of '-progress  employed.  ,  If  a  choice  is  to  be  made  among 
the  various  measures,  amount  scores  will  prove  more 
serviceable  because  of  their  wider  applicability.  In  a 
"speed"  test,  amount  scores  can  be  employed  inter- 
changeably with  time  scores.  In  a  "power"  test,  however, 
in  which  the  items  are  arranged  in  an  order  of  progres- 
sively increasing  difficulty,  a  time  score  would  be  meaning- 
less. If,  for  example,  in  a  3O-minute  test  consisting  of 
10  problems,  all  the  subjects  attempt  all  the  problems 
but  the  number  of  correct  solutions  ranges  from  I  to  10, 
it  would  be  absurd  to  report  that  the  average  time  per 
problem  ranged  from  30  minutes  to  3  minutes.  The  better 
subjects  worked  no  faster  than  the  poorer  subjects  since 
all  members  of  the  group  tackled  all  the  problems. 

'  the  scores  have  been  found,  there 
irmmmt, a£  Jfe .  data.  A jivefy 

.  goiug  on  regarding  tfee  use  of  rela- 
tive or  absolute  pleasures  of  variability  in  practice  ex- 


periments.  When^absplute  measures  are  used,  such  as  the 
standard  deviation^  or  gross  gains  made  by  Initially  high 
and  low  individuals  or  groups,  variability  tends  to  increase 
with  practice.  When,  on  the  other  hand,  relative  measures 
are  employed,  such  as  the  coefficient  of  relative  variabil- 
ity,1 or  some  measure  based  upon  relative  or  percentage 
gains,  then  variability  decreases  with  practice  in  most 
cases.  The  fundamental  objection  against  the  use  of 
relative  measures  has  already  been  discussed  in  a  previous 
chapter  (cf.  Ch.  II).  It  was  there  demonstrated  that, 
sig.ce  scores  on  most  current  psychological  tests  are  not 
measured  from  an  absolute  zero  point  of  performance, 
any  ratios  or  quotients  computed  with  such  scores  may 
be  entirely  misleading;  the  addition  of  a  few  easy  items 
at,,,the  .lower  end  of  the  scale  might  completely,  reverse 
the  relationship  between  the  obtained  values. 

Thus  it  would  seem  that  absolute  measures  of  variabil- 
ity are  preferable  for  a  purely  negative  reason,  if  for  no 
other.  Since  relative  measures  are  ruled  out  by  the  use 
of  arbitrary  starting  points  in  the  tests,  no  alternative 
is  left.  We  may,  however,  inquire  more  directly  into  the 
logic  of  using  absolute  or  relative  measures  in  practice 
experiments.  Tjh^e  argyineatiii,  support  of  relative,  meas- 
ures is  that,  since  the  numerical  size  of  scores  changes 
Ln,tb&.£0ur$?  pi, practice,,  the  scqres,  ai-e  ixo.t.  expressed  in 
the  same  units  throughout  4&d  ,heace,ab?plute,,measures 
will  not  be  comparable  fromrtr;i^lTtQytnal.  Through  chance 
alone^,  the, argument  runs,  absolute  variability, will, mc%2ase 
wJiea- the^sizfiMjoL^  jwhen  „  the 

scores  decrease,  such  changes,  being  therefore  of  the  nature 
of  a  statistical  artifact. 

It  is  perfectly  true  that,   other  things   being  -  equal, 



numerically  larger  scores  will  exhibit  greater  variability. 
Obviously,  if  the  standard  deviation  o£  a  distribution 
of  time  scores  is  10  minutes,  the  standard  deviation  of  the 
same  scores  expressed  in  seconds  will  be  600.  For  the 
same  reason,  the  standard  deviation  of  the  number  of  A's 
cancelled  in  one  minute  cannot  be  compared  directly 
with  that  of  the  number  of  additions  performed  during  an 
equal  period,  since  the  latter  scores  would  be  much  smaller.1 
This  type  of  argument  does  not  necessarily  hold,  how- 
ever, when  the  same  test  is  given  to  different  groups  or 
to  the  same  group  under  different  conditions,  such  as 
before  and  after  practice.  Let  us  suppose  that  the  average 
score  of  a  certain  group  I  on  an  intelligence  test  is  25  points 
and  that  of  group  II  50  points.  It  does  not  necessarily 
follow  from  this  difference  in  averages  that  group  II 
will  have  a  larger  standard  deviation  than  group  I.  In 
fact,  the  opposite  might  very  likely  be  the  case.  If 
group  I,  for  example,  consisted  of  unselected  third  grade 
public  school  children  and  group  II  of  superior  sixth 
grade  children  in  a  private  school,  the  latter  would  prob- 
ably have  a  lower  standard  deviation.  Similarly,  one  could 
assemble  without  too  much  difficulty  two  groups  of  men, 
whose  average  heights  were  64  and  72  inches  respectively, 
but  whose  standard  deviation  was  8  in  both  cases.  It 
would  be  quite  absurd  to  insist  that  the  taller  group  is 
"actually"  less  variable  in  height  than  the  shorter,  or  to 
suggest  that  the  inches  used  in  measuring  height  had  in 
some  mysterious  fashion  changed  in  value  from  group  I 

to  group  II.  I\J52HlfLjS2Hi .  ^at  ^  ,?9mParJs!P£.J2f 
variability  before  and  after  practice  is  more  similar  in 

pr^iplt'to  t^tjsibov^  examples  tj^fl to  jthe  raeasuWTOSPt 
of  variability  in  diffe^nt  te,^^,  Tlie, ,jz§£,  of  absolute 
measures  in  this  connection  is  therefore^stifiable. 

1  Cf.  Ch.  II  for  a  fuller  discussion. 




If  the  problem  be  formulated  in  specific  terms,  we  can 
legitimately  inquire  into  the  effects  of  practice  upon  mdi- 
vidual  differences.  For  reasons  of  convenience,  as  discussed 
above,  we  "may  arbitrarily  define  equal  practice  as  equal 
time  spent  in  practice  and  express  scores  in  terms  of 
amount  done  per  unit  time.  Variability  is  to  be  measured 
in  absolute  terms,  for  the  reasons  stated  in  the  preceding 



(After  Anastasi,  I,  pp.  40-42) 





Hidden  Words 



















































































































































































1  The  scores  on  all  of  these  tests  were  transmuted  into  an  equal  unit  scale  and 
are  thus  directly  comparable  from  one  test  to  the  other. 


paragraph.  Under  these  conditions,  the  findings  of  differ- 
ent experimenters  agree  very  closely.  An  investigation 
by  Anastasi  (i)  illustrates  the  general  methods  and  find- 
ings of  such  studies. 

Four  groups,  each  comprising  from  114  to  200  college 
students,  were  given  continuous  practice  in  one  of  four 
tests.  The  tests  included  A~Cancellation,  Hidden  Words,1 
Symbol-Digit  Code  Learning,  and  Vocabulary  Learning.2 
The  practice  consisted  of  15  4-minute  trials  in  Hidden 
Words  and  20  2-minute  trials  in  each  of  the  other  tests, 
a  different  group  of  subjects  being  employed  for  each 
test.  The  average  scores  and  standard  deviations  of  the 
scores  on  each  trial  are  reproduced  in  Table  VI. 

It  will  be  readily  seen  that,  the  standard  deviation^  rise 
with  practice  in,  every  test.  It  was  also  found  that  indi- 
viduals tend  to  maintain  the  same  relative,,  staading  in 
the  group  in  the  course  of  practice,  the  correlations  be- 
tween initial  and  final  scores  of  the  same,  sahjecta.  being 
consistently  positive  and  usually  high.  For  the  four  tests, 
these  correlations  were: 

Cancellation  .6725 

Symbol-Digit  .2981 

Vocabulary  -S°73 

Hidden  Words  .8239 

Such  correlations  indicate  a  tendency  for  the  individual 
who  is  best  in  the  group  at  the  outset  to  remain  at  the 
top  after  practice,  for  the  one  who  is  lowest  to  remain 
at  the  bottom,  and  so  on.  This  is  commonly  found  to  be 
the  case  in  all  experiments  on  practice. 

Both  the  tendency  to  maintain  the  same  relative  posi- 
tion during  practice  and  the  increase  in  absolute  variabil- 

1  Subjects  were  to  underline  all  four-letter  English  words  which  were  "hidden" 
in  a  page  of  pied  type. 

2  Subjects  learned,  by  the  method  of  paired  associates,  a  "vocabulary"  of 
nonsense  syllables;  the  test  is  similar  to  code  learning,  but  more  difficult. 



ity  are  illustrated  graphically  in  Figure  25.  This  shows  the 
learning  curves  of  two  subjects  on  the  Hidden  Words 
test.  The  subjects  were  selected  near  the  extremes  of  the 
distribution,  the  differences  between  their  initial  scores 

95  r- 



I  70 


I  60 









I I 

I I 




-I  95 



10     11     12     13     14     15 

WITH  PRACTICE.  (Unpubl.  data  from,  investigation  of  Anastasi,  I.) 

being  very  large.  It  will  be  noted  that  the  curves  do  not 
at  any  time  cross  and  that  they  diverge  with  practice, 
the  differences  between  the  two  individuals  being  much 
larger  on  the  fifteenth  trial  than  they  were  on  the  first. 
Itjias ^be^ii, frequently  assumed-  that  ifJiidisdduaLdiffer- 
s Jua  performance  incuase  with  practice, ,  J:Jxc^jC£ja,,  be 

to  hereditary  differ,erj,ce$^  wh 
se^  .they  inust  have  t resulted^  in ,  J^ 


from  inequalities  of  past  training  and  environmental  stim-^ 
ufation.  It  is^argued  that  .when  subjects  undergo  a,  pro- 
longed period  of  equal  training,  the  differences  in  their 
past  experience  with  the  given  task  are  thereby  wiped  out. 
This  assertion  is  open  to  question.  The  influence  of  en- 
vironmental factors  upon  the  development  of  the  indi- 
vidual is  ordinarily  cumulative.  If  one  individual's  past 
experience  has  made  him  more  proficient  than  another 
in  a  certain  task,  we  should  expect  him  to  be  better  fitted 
to  profit  from  instruction  for  that  very  reason.  Suscepti- 
bility to  training  can  itself  be  environmentally  deter- 
mined, and  if  so  determined  there  is  no  reason  to  assume 
that  it  will  disappear  with  additional  training.  The  indi- 
vidual who  has  been  handicapped  by  a  poor  environment 
may  lack  the  necessary  intellectual  tools  to  profit  from 
instruction.  Thus,  had  the  Wild  Boy  of  Aveyron  (cf. 
Ch.  Ill)  and  a  boy  of  the  same  age  from  a  middle  class 
English  home  been  put  through  an  identical  one-year 
course  in  the  reading  of  French,  the  differences  in  their 
abilities  to  read  that  language  would  have  been  far  greater 
at  the  end  of  the  course  than  at  the  beginning.  ,  It  i$ 
obviously  unnecessary  to  assume  an  hereditary  basis^of 
individual  differences  in  order  to  account  for  the  increase 
in  variability  in  such  an^,  example.  The  more  the  indi- 
vidual has  learned  in  the  past,  the  more  he  will  be  able 
to  learn  in  the  present.  To  use  a  rather  crude  analogy, 
we  might  say  that  practice  does  not  add  to  the  individual's 
ability,  but  multiplies  it. 


1.  Anastasi,  A.   "Practice  and  Variability:  a  Study  in  Psycho- 
logical Method/'  Psychol.  Man.,  1934,  45.  Pp.  55. 

2.  Carmichael,  L.     "A  Further  Experimental  Study  of  the 
Development  of  Behavior,"  Psychol.  Rev.,  1928, 35, 253-260, 


3.  Casey,  M.  L.,  Davidson,  H.  P.,  and  Harter,  D.  I.    "Three 
Studies  on  the  Effect  of  Training  in  Similar  and  Identical 
Material  upon  Stanford-Binet  Test  Scores,"  2jth  Yearbook, 
Nat.  Soc.  Stud.  Educ,,  1928,  Part  I,  431-439. 

4.  Dennis,  W.    "The  Effect  of  Restricted  Practice  upon  the 
Reaching,  Sitting,  and  Standing  of  Two  Infants,"  /.  Genet. 
PsychoL,  1935,  47,  17-32. 

5.  Gates,  A.  I.    "The  Nature  and  Limit  of  Improvement  Due 
to  Training,"  27th  Yearbook,,  Nat.  Soc.  Stud.  Educ.,  1928, 
Part  I,  441-460. 

6.  Gesell,  A.    Infancy  and  Human  Growth.   N.  Y.:  Macmillan, 
1928.   Pp.  418. 

7.  .    "The  Individual  in  Infancy."     Ch.  16  in  The  Foun- 
dations  of  Experimental    Psychology,    C.    Murchison,    ed. 
Worcester:  Clark  Univ.  Press,  1929.  Pp.  907. 

8.  Greene,  K.  B.    "The  Influence  of  Specialized  Training  on 
Tests  of  General  Intelligence,"  27th   Yearbook,  Nat.   Soc. 
Stud.  Educ.,  1928,  Part  I,  421-428. 

9.  Kincaid,  M.    "A  Study  of  Individual  Differences  in  Learn- 
ing," Psychol.  Rev.,  1925,  32,  34-53. 

10.  McGraw,  M.  B.    Growth:  a  Study  of  Johnny  and  Jimmy, 
N.  Y.:  Appleton-Century,  1935.  Pp.  319. 

11.  Peterson,  J.,  and  Barlow,  M.  C.  "The  Effects  of  Practice  on 
Individual   Differences,"   27th    Yearbook,   Nat.    Soc.    Stud. 
Educ.,  1928,  Part  II,  211-230. 

12.  Reed,  H.  B.     "The  Influence  of  Training  on  Changes  in 
Variability   in   Achievement,"    Psychol.    Mon.,    1931,     41. 

PP.  59. 

13.  Spalding,  D.  A.    "Instinct:  with  Original  Observations  on 
Young  Animals,"  Pop.  Sci.  Mo.,  1902,  61,  126-142. 


e  subsecluently  shown,  the  distinction  between 
investigations  of  training  and  those  of  mental  growth  is  a 
superficial  one.  It  is  only  for  convenience,  therefore,  that 
the  former  were  discussed  in  the  preceding  chapter,  while 
the  latter  will  be  surveyed  in  the  present  one.  The  data 
of  the  two  chapters  should  be  considered  as  a  whole.  A 
few  studies  are  difficult  to  classify  into  one  or  the  other 
category;  this  is  especially  true  of  experiments  on  very 
young  children,  such  as  those  reported  in  Chapter  VI. 
For  the  purposes  of  the  present  discussion,  however,  we 
may  regard  studies  of  "growth"  as  those  in  which  mental 
progress  at  successive  ages  is  observed  and  charted,  with 
no  attempt  to  alter  the  normal  course  of  development. 


Growth  curves  were  first  plotted  to  show  the  develop- 
ment of  physical  traits,  such  as  height,  weight,  bodily 
proportions  as  indicated  by  various  indices,  and  the  like. 
An  example  of  such  a  curve,  showing  the  changes  in  height 
in  groups  of  tall,  average,  and  short  girls  between  the 
ages  of  5  and  17,  is  given  in  Figure  26.  As  a  descriptive 
technique  for  portraying  more  vividly  the  course  of  de- 
velopment of  structural  characteristics,  the  growth  curve 
has  proved  serviceable  and  Intelligible.  The  physical 
data  are  relatively  easy  to  interpret  and  unambiguous. 


es  pf  "mental  gjpwtji,"  a 
additional  confusion  into  an  already  difficult  problem 




At^best  these  curves  are  only  a  descriptive  summary  of 
changes  produced  by  a  multiplicity  of  factors;  by  lumping 
all  such  factors  together  and  giving  them  a  semblance  of 
systematic  growth,  the  main  issues  are  only  obscured. 

Two  general  methods  have  been  followed  in  the  study 
of  age  changes  in  mental  traits.    The  more  direct  ap- 


1 145 

1 140 



f  125 






5    6     7    8     9    10  11  12  13  14   15  16  17  18 
Age  in  Years 

(After  Baldwin  and  Stecher,  i,  p.  13,) 

proach  is  to  retest  a  group  of  individuals  at  successive 
ages,  a  very  laborious  and  time-consuming  procedure.  This 
method  is  also  open  to  the  objection  that  specific  practice 
effects  may  operate  in  successive  administrations  of  the 
same  test  or  closely  parallel  forms.  The  more  common 
method  is  to  test  different  age  groups  at  the.  same  time. 
The  average  score  of  each  age  group  is  th^n  regarded  as 
indicative  of  the  normal  course  of  development  at  suc- 
cessive age  levels.  Such  a  procedure  is  subject  to  serious 
selective  factors,  since  the  different  age  groups  may  not 



be  comparable.     An  illustrative  study  by  each  method' 
will  be  described  briefly. 

In  one  of  the  most  extensive  investigations  of  "  mental 
growth,"JTeagarden  (13)  tested  408  subjects  between  the 
ages  of  I2§  an3TT67"The  standing  of  the  group  as  a  whole 
on  intelligence  tests  showed  them  to  be  typical  of  Amer- 
ican children  and  adolescents,  their  median  Stanford- 
Binet  I.Q.  being  93,5,  with  a  range  from  61  to  136. 
All  subjects  were  given  the  Stanford-Binet,  Army  Alpha, 
Pressey  Senior  Classification  1  and  Stenquist  Mechanical 
Aptitude  Tests.  Growth  curves  were  plotted  for  each  test. 
Apart  from  many  irregularities  resulting  from  selective 
factors,  as  well  as  certain  differences  from  test  to  test, 

the  typ 

A  compos- 
ite curve  for  the  four  tests  is  reproduced  in  Figure  27. 

65  r 

|  55 

|  45 



14        15        16        17        18        19      -20 
Chronological  Age 

(After  Teagarden,  13,  p.  78.) 

^  in-  a  ^ 

differences  increase3"with  age,  variability'being 

""  "*"  '  ""  "  ^^"r 


Baldwin  and  Stecher  (i)   administered  the  Stanford- 
to^TgroupoFTt43   normal  and  superior  children 

1  A  group  test  of  "general  intelligence.'1         .  '  , 



between  the  ages  of  5  and  14.  The  study  differed  from 
Teagarden's  In  many  respects.  The  group  was  younger 
and  included  more  children  at  the  higher  intellectual 



8       9      10     11     12     13     14 

Chronological  Age  in  Years 

STANFORD-BINET  MENTAL  AGES.    (After  Baldwin  and  Stecher,  i,  p.  11.) 

levels,  the  LQ.Y  ranging  from  90  to  167.  The  retest 
technique  was  employed,  each  child  being  tested  from 
two  to  five  times  within  a  four-year  period.  Growth 
curves  were  plotted  exclusively  in  terms  of  mental  ages. 


In  constructing  these  curves,  the  retest  data  were  com- 
bined with  data  on  equivalent  groups  of  children  at 
different  age  levels,  so  that  the  curves  are  not  based 
on  the  same  subjects  throughout  the  age  range.  In  Fig- 
ure 28  will  be  found  the  growth  curves,  plotted  separately 
for  normal  and  superior  boys  and  girls.  It  will  be  seen 
that  these  curves  resemble  straight  lines  very  closely  and 
do  not  exhibit  the  usual  negative  acceleration.  /The 
curves  of  the  normal  and  superior  groups  diverge  with  age, 
again  indicating  an  increase  in  variability  with  age. 

The  problem  of  the  effect  of  age  upon  the  extent  of 
individual  differences  is  subject  to  the  same  difficulties 
as  that  of  practice  and  variability.  Since  most  studies 
have  employed  amount  scores,  absolute  variability  has 
usually  been  found  to  increase  with  age,  as  it  does  with 
training.  In  the  interpretation  of  the  growth  curve  itself 
as  a  picture  of  the  course  of  mental  development,  addi- 
tional and  more  serious  difficulties  are  met.  The  weak- 
nesses inherent  in  both  the  retest  method  and  the  equiva- 
lent group  method  have  already  been  mentioned.  It  is 
the  latter  and  cruder  method  which  has  more  frequently 
been  followed  for  reasons  of  convenience.  The  choice 
of  subjects  also  affects  the  type  of  curve  obtained.  When 
only  younger  children  are  tested,  for  example,  the  curve 
may  resemble  a  straight  line  because  it  represents  only 
the  first  part  of  the  growth  curve  and  is  cut  off  before 
the  negative  acceleration  becomes  sufficiently  large  to  be 

xpjnplex  scale  such  as tj^e ,  ..SJtMlojd-Binet  is 
g  same  abilities  are  not  measured  throughout 
the  age  range.  ^At  the  ^  upper  age  Jevel^ 
scaLsJ^jiiu^^mpre  heavily  loaded  ^itjb^yjgrbajj^ts.  It 
has  also  been  pointed  out  that  the  form  of  the  curve  of 
mental  growth  may  differ  with  the  difficulty  of  the  test. 



In  a  relatively  easy  .task,,  performance  will  improve  rap- 
Myjdunng  '"the  first  few  years  and  more  slowly  later  on 
as^a  perfect  score  is  approached.  In  ajrelatively  difficult 
task,  on  the,,  other -hand,  or  a  task  which  requires  a  certain 
degree  of  general  information  or  mastery  of  techniques 
^forefit  can  be  properly  executed,  progress  will  be  slow 
at  first  and  much  more  rapid  at  the  upper  age  levels. 
The  latter  task  would  thus  give  a  positively  rather  than  a 
negatively  accelerated  curve.  This  has  been  illustrated 
by  Freeman  (5)  by  means  of  performance  curves  of  suc- 
cessive school  grades  on  easy  and  difficult  sentences  in 
the  Trabue  Sentence  Completion  Scale.  The  curves  for 
five  representative  sentences  of  different  degrees  of  diffi- 
culty are  shown  in  Figure  29. 

100  r 


II       III      IV       V       VI     VII    VIII 
School  Grade 

OF  VARYING  DIFFICULTY.   (After  Freeman,  5,  p.  336.) 



curve  is  to  be  found  in  the_unjt^^mwterms  of  which  the 
cs^v^^piQ^dJ''>>TE^'use  of  mental  ages,  as,  in  the  study 
by  Baldwin  and  Stecher  reported  above,  is  particularly 
misleading.  It  can  be  readily  demonstrated  that  if  aver- 
age mental  age  is  plotted  against  chronological  age,  the 
result  must  be  a  straight  line,  unless  the  test  is  not  suf- 
ficiently well  standardized  or  is  unsuited  to  the  group 


upon  which  It  was  employed.  To  obtain  a  mental  age 
growth  curve  which  approximates  a  straight  line  simply 
serves  to  show  that  the  test  fulfills  its  purpose  satis- 
factorily, since  an  age  scale  is  constructed  in  such  a  way 
that  the  average  child  will  progress  one  year  in  mental 
age  during  each  year  of  life.  Thus  the  successive  mental 
age  units  are  adjusted  so  as  to  rule  out  automatically  any 
differences  in  the  amount  of  improvement  from  year  to 
year  and  are  completely  unsuited  to  an  analysis  of  the 
course  of  mental  development.  The  Jinits  in  which  a 
growth  curve  isjglotted  should  be  equal  ^EroupioufTiLe 
range;  otherwise  they  will  present  a  completely  distorted 
picture.  Various  statistical  devices  have  been  suggested 
in  order  to  arrive  at  a  "true"  picture  of  the  course  of 
mental  development,  but  these  techniques  are  still  im- 
mersed in  much  technical  controversy.1 

ItJsjL2£arent  that  for  many  reasons  the  available  curves 
of  mental  growth  are  of  doubtful  significance.  Apart  from 
technical  difficulties,  however,  we^may  question  even  the 
concept  of  a  growth  curve  in  the_  analysis ,  of  mental 
development.  What  such  .a.  curve,  actually .  shows  is  the 
performance  of  the  individual  at  different  ages  in  some 

1  Thorndike  (15,  pp.  463-466)  presents  a  negatively  accelerated  curve  to  in- 
dicate the  general  trend  of  development  in  scores  on  the  CAVD  Intelligence 
Examination,  which  is  scaled  in  equal  units  from  approximately  absolute  zero. 
This  curve,  however,  represents  a  very  rough  estimate  based  on  data  from  many 
widely  varying  groups  of  subjects  who  had  taken  different  levels  of  the  test. 

Thurstone  (17)  plotted  the  Stanford-Binet  scores  of  4208  children  between 
the  ages  of  3  and  17,  in  terms  of  "absolute  scale  units,"  a  scaling  technique  which 
he  had  devised  particularly  for  the  study  of  mental  growth.  The  curve  he  ob- 
tained was  positively  accelerated  at  the  lower  age  levels  and  negatively  accelerated 
at  the  upper. 

Courtis  (3)  has  suggested  a  unit  which  he  calls  the  isochron  for  the  measure- 
ment of  growth.  He  defines  an  isochron  as  one  one-hundredth  of  the  total  time 
required  for  maturation.  In  plotting  a  growth  curve,  he  takes  the  developments 
made  in  equal  intervals  of  the  total  maturation  period  as  equal.  When  isochrons 
are  substituted  for  conventional  units  of  achievement  and  plotted  against  ago. 
a  straight  line  is  obtained  as  the  growth  curve. 


standard  test  situation.  Such  a  curve  does  not  differ 
in  any  essential  respect  from  a  learning  curve.  In  both 
cases,  the  subject  is  tested  under  similar  conditions  at 
successive  intervals  and  his  progress  is  charted  on  the 
curve.  Learning  curves,  to  be  sure,  usually  cover  a  shorter 
period  of  time  than  growth  curves,  although  a  practice 
experiment  could  conceivably  extend  over  several  years. 
The  major  difference  between  learning  curves  and  growth 
curves  seems  to  be  that  in  the  former  the  subject  is  given 
special  training  under  rigidly  controlled  experimental  con- 
ditions, while  in  the  latter  he  is  left  to  his  own  resources. 
Thus  jjLwpuld^seem  that  a  mental  growth  curve  is  at  best 
a  practice  curve  obtained  in  the  absence  of  controlled 
conditions.1  It  reflects  the  cumulative  effects  of  the  ran- 
dom training  and  experience  of  everyday  life,  without 
adding  anything  essentially  new  to  the  picture. 

It  follows  from  this  discussion  that  growth  curves  are 
specific  to  the  cultural  milieu  in  which  they  are  obtained. 
If  the  learning  conditions  differ  from  one  group  to  an- 
other, the  curves  of  mental  growth  should  likewise  be 
expected  to  differ.  It  might  be  suggested  that  the  mental 
growth  curve  could  still  serve  a  useful  purpose  as  a  descrip- 
tive device.  As  such  it  would  indicate  the  general  course 
of  development  to  be  expected  under  given  cultural  con- 
ditions^ and  would  characterize  individuals  of  different 
age  levels  within  a  specific  group.  For  this  purpose,  how- 
ever, it  seems  that  a  more  useful  and  intelligible  picture 
could  be  obtained  by  observing  cross-sections  2  of  the 

1  This  seems  to  be  in  general  agreement  with  the  view  expressed  by  Courtis 
(3,  4),  who  regards  growth  as  development  under  constant  environmental  condi- 
tions, and  subsumes  practice  curves  under  the  heading  of  growth  curves.    It 
seems  to  the  writer,  however,  that  the  issues  involved  would  be  better  clarified 
if  growth  curves  were  regarded  as  a  type  of  practice  curve,  rather  than  vice 

2  Cf.,  e.g.,  the  interesting  descriptions  of  the  individual  at  various  ages  given 
by  Hollingworth  (7). 






w  Wi 

n  5  £ 

1-3      S?     O 



a  w 







0     (5 



J  ON  O  *^  t^oo  "-o  en  enVO  w 

M   cn        w   ^  t^-oo  MD  i-o  IN.  M  10  u->  O^O  00  Jt>.  c 


M  in  ^vo  VOOD 

w  «  -rh  co  IN 

ON  ^ 

O  o 

O  CO 

M     M  °°^ 

v    0 

O\  en 

to  O 
cr>  oV 

ON  tn 

to  c 

oo  en 

u-)  O 
Q  «^ 

NO      « 

V£)   en 


O  ONOO  «>-vo  to  ^t-  en  M  M  O  a\oo  t>.vo  UTI  ^  to  c*  «    I     JJ  r/ 




individual's   behavior   at  various   intervals,    rather  than 
tracing  isolated  traits  through  successive  years. 


Lively  discussion  has  been  stimulated  by  the  question 
of  intellectual  maturity  and  decline.  Popular  indignation 
or  alarm  was  aroused  when,  on  the  basis  of  the  army 
testing  data,  the  report  was  voiced  about  that  the  aver- 
age American  male  had  the  mentality  of  a  14-year-old 
child.  This  simply  meant,  of  course,  that  the  comparison 
of  Stanford-Binet  and  Army  Alpha  scores  of  a  representa- 
tive group  of  soldiers  in  the  United  States  army  indicated 
that  the  average  individual  does  not  improve  beyond  the 
age  of  14  in  performance  on  common  intelligence  tests. 
These  findings  have  subsequently  been  questioned,  how- 
ever^jiiore  recent  investigations  ,,,.such..  as  those  of  Tea- 
garden  (13)  and  Thorndike  (14)  haying  placed  the  "limit 
of  intellectual  growth"  in  the  late  teens  or  early  twenties. 
In  general,  mental  development  seemed  also  to  continue 

longer  in  subjects  of 
higher  educational 

A  closely  related 
question  concerns  the 
decline  of  mental  ac- 
tivity. Jones  and 
Conrad  (8)  gave  the 
Army  Alpha  to  1191 
individuals  between 
the  ages  of  10  and 
60,  living  in  19  vil- 
lages in  rural  sec- 







30          40 

Chronological  Age 



NEW  ENGLAND  GROUPS.  (After  Jones  and  Con- 
rad, 8,  p.  241.) 

tions  of  New  England.  In  Table  VII  are  shown  frequency 
distributions  as  well  as  averages  and  standard  deviations  of 


successive  age  groups.  The  age  changes  in  average  score  are 
shown  graphically  in  Figure  30.  It  will  be  seen  that  the  aver- 
age score  rises  from  44.4  at  age  10  to  100.7  at  19-21.  Be- 
yond this  there  appear  irregular  and  small  drops.  Even  for 
the  oldest  group  tested,  the  drop  is  not  very  large.  Selective 
factors  may  account  in  part  for  these  findings.  Among  the 
older  groups  there  were  more  individuals  who  pleaded  ex- 
emption from  the  test  because  of  failing  eyesight,  difficulty 
in  reading,  and  similar  reasons.  The  less  intelligent  and 
less  energetic  older  subjects,  furthermore,  were  less  likely 
to  come  to  the  community  centers  where  the  tests  were 
given,  as  was  evidenced  by  the  fact  that  subjects  tested 
in  their  own  homes  showed  a  much  sharper  age  decline 
than  those  tested  in  community  centers.  An  analysis  of 
age  curves  of  the  different  tests  in  Army  Alpha  revealed 
fundamental  differences  from  test  to  test,  the  curves  for 
"naming  opposites"  and  " general  information,"  for  ex- 
ample, showing  no  decline  with  age. 

In  regard  to  variability,  it  should  be  noted  that  indi- 
vidual differences  were  large  at  all  ages,  as  shown  both 
by  the  distributions  and  by  the  size  of  the  standard 
deviations.  Overlapping  of  different  age  groups  is  very 
marked,  the  range  of  scores  within  any  one  age  group 
being  much  greater  than  the  largest  average  difference 
between  groups.  Finally,  the  standard  deviation  shows  a 
fairly  clear-cut  tendency  to  rise  with  age,  even  though 
selective  factors  would  tend  to  make  the  older  groups 
tested  more  homogeneous  than  the  younger.  It  is  inter- 
esting to  find  that  the  standard  deviations  of  the  older 
groups  are  larger  than  those  of  the  younger  in  spite  of 
the  fact  that  the  averages  decrease  with  age.  For  exam- 
ple, the  50-54  year  group  has  an  average  of  81.3,  slightly 
lower  than  that  of  the  15  year  group  which  is  85.7.  The 
standard  deviation  of  the  former,  however,  is  43.1  as 


compared  to  32.3  for  the  younger  group.  Thus  the  changes 
in  .standard,  deviation  with  age  seem  to  be  something 
more  than  a ,  statistical  artifact  which  might  result  from 
the.  changes.  In  size  of  scores.  As  the  individual  has  more 
experiences  and  is  subjected  to  a  wider  variety  of  environ- 
mental stimulation  with  increasing  age,  more  possible 
sources  of  variation  are  thereby  introduced  and  individual 
differences  will  continue  to  increase  indefinitely. 

Willoughby  (19),  in  his  investigation  of  family  resem- 
blances in  mental  traits,  also  compared  averages  and 
standard  deviations  on  each  of  II  tests1  for  successive 
age  groups  from  childhood  to  60  years.  The  scores  again 
rose  until  the  late  teens  or  early  twenties.  Beyond  that 
age,  however,  the  different  tests  presented  somewhat  vary- 
ing pictures.  A  marked  decline  in  score  was  found  in  tests 
of  series  completion,  analogies,  opposites,  substitution 
learning,  history-literature  information,  and  making  com- 
parisons. In  arithmetic  reasoning,  on  the  other  hand, 
no  decline  was  exhibited.  The  latter  task  is  one  which  is 
more  useful  and  more  frequently  met  in  everyday  life. 
The  decline  of  the  older  groups  in  most  tests  which  do 
not  enter  into  ordinary  adult  activities  could  be  attrib- 
uted to  the  fact  that  the  formal  education  of  the  older 
subjects  is  more  remote  and  probably  less  in  amount.  The 
general  level  of  education  has  increased  phenomenally 
during  the  past  50  years,  a  fact  which  cannot  be  ignored 
in  comparing  older  and  younger  groups. 

Miles  and  Miles  (9)  conducted  a  comprehensive  survey 
of  age  changes  in '  intelligence  test  scores.  A  total  of 
823  persons  ranging  in  age  from  7  to  94  were  given  a 
special  adaptation 2  of  the  Otis  Self-Administering  Intelli- 

1  See  Ch.  IV  for  fuller  description. 

2  Consisting  of  the  first  60  items  of  the  test,  given  with  a  1 5-minute  time  limit. 
The  scores  so  obtained,  however,  were  found  to  correlate  highly  with  scores  on 
the  30-minute  test,  on  various  groups. 


gence  Test.  The  averages  and  standard  deviations  shown 
in  Table  VIII  are  based  on  617  adults  of  both  sexes  tested 
in  one  city  and  are  typical  of  the  general  findings.  The 
averages  show  practically  no  change  between  15  and  29; 
between  30  and  50  there  are  very  small  but  consistent 
decreases;  beyond  50  the  decline  is  more  rapid  and  con- 
tinues up  to  the  highest  age  group  tested.  Variability 
indicates  no  consistent  change  with  age  unless  it  be  a 
slight  increase  during  the  middle  period  between  40  ar^d 
50,  as  compared  to  the  youngest  and  oldest  groups.  This 
is  in  general  agreement  with  the  findings  of  other  studies 
reported  above.  Wide  individual  differences  were  ob- 
served within  all  age  groups,  with  marked  overlapping. 


(After  Miles  and  Miles,  9,  p.  53) 


of  Cases 




of  Cases 


















ii.  16 
















































The  correlations  between  each  individual's  score  and 
his  age  brought  out  the  same  trends  suggested  by  the 
comparison  of  averages.  The  correlation  throughout  the 
entire  adult  age  range,  from  20  to  95,  was  approximately 
—  .50,  showing  a  clear-cut  tendency  for  the  older  persons 
to  make  lower  scores.1  When  only  individuals  between 

1  Similar  results  have  recently  been  obtained  in  a  study  on  a  small  group  of 
"mentally  normal"  hospital  patients  who  were  submitted  to  an  intensive  testing 
program  (18). 






15  - 

15  and  55  are  included,  the  correlation  is  —.283;  when 
Dnly  the  older  subjects  between  50  and  95  are  employed, 
the  correlation  rises  to  —.372.  This  suggests  a  more  rapid 

rate  of  decline  during  the 
upper  ages,  thus  corrobo- 
rating the  findings  on  aver- 
ages. The  fact  that  all  the 
correlations  are  far  from 
a  perfect  —  i.oo  is  attrib- 
utable to  the  wide  individ- 
ual variation  within  age 
groups  and  the  overlap- 
ping of  different  age 

As  a  further  analysis 
of  age  changes,  the  sub- 
jects were  classified  into 
four  groups  in  respect  to 
the  amount  of  formal 
education  which  they  had 
received.  In  Figure  31 
will  be  found  the  four 
curves,  A  to  D,  showing 
the  average  score  of  each 

educational  group  at  successive  decades  between  20  and  80. 
The  groups  were  constituted  as  follows: 

Group  A— four  years  of  college  plus  additional  grad- 
uate or  professional  training 

Group  B — one  year  of  college  and  higher  (includes 
group  A) 

Group  C — one  to  four  years  of  high  school  or  its  equiv- 

Group  D— eight  grades  of  elementary  school  or 


Beyond  College 

College  and  Beyond  (1-10 

years;  includes  the  above) 
High  School  or  Equivalent  (1-4  years) '' 

-Grade  School  (0-8  years) 

i 1         i         i 

20       30      40       50      60       70       80 
Chronological  Age 


MENTAL  GROWTH  .    175 

Although  all  groups  show  a  decline  from  the  lower  to 
the^ggpTer  "age  levels,  this  decline  is  smaller  for  the  higher 
educational  Jgtcmjts*  The  curves  do  not  cross,  individuals 
in  the  higher  educational  groups  obtaining  higher  average 
scores  than  those  in  the  lower  groups  at  all  ages.  It  is 
also  interesting  to  note  that  the  lowest  point  on  curve  A, 
reached  by  the  70  year  group  is  still  higher  than  the 
highest  points  of  curves  C  and  D.  TJIUSLJ*  yo-year-old 
person  .who  had  pursued  at  least  one  year  of  graduate 
work  would  be  expected  to  score  higher  than  a  2O-year- 
613  elementary  ,or  High  school  graduate.  The  significance 
of  this  finding  will  be  elaborated  below,  in  conjunction 
with  other  data. 


In  a  series  of  varied  and  extensive  experiments  on  adult 
learning,  Thorndike  (16)  found  no  appreciable, decline  in 
this  ability  between  the  ages  of  20  and  45.  Several  groups 
of  adults  were  put  through  a  long  period  of  training  in 
different  tasks  and  their  progress  was  observed.  The 
groups  varied  from  an  intellectually  and  educationally 
superior  group  of  graduate  students,  through  more  nearly 
average  subjects  attending  evening  high  schools  and  secre- 
tarial schools,  to  a  somewhat  inferior  group  of  Sing  Sing 
prisoners.  The  major  findings  have  been  summarized  in 
Table  IX.  In  each  task,  the  gain  through  practice  made 
by  an  older  group  (35  or  over)  has  been  expressed  as  a 
percent  of  the  gain  made  by  a  younger  group  (20-24) 
under  the  same  conditions.  Thus,  if  the  percentage  is  100 
it  indicates  that  the  two  groups  made  equal  gains;  if  it  is 
less  than  100,  the  younger  group  gained  more;  if  over  100, 
the  older  gained  more. 

1  For  a  comprehensive  survey  of  studies  on  adult  learning,  the  reader  is  re- 
ferred toRuch  (n). 



It  is  apparent  from  the  data  of  Table  IX  that  age 
differences  in  favor  of  the  younger  group  are  larger  in  the 
more  meaningless  rote  learning  tests  and  are  absent  or 
reversed"  in  the  more  practically  useful  and  meaningful 
functions.  In  most  tasks,  the  older  person  could  compen- 
sate for  any  deficiency  in  learning  capacity  by  greater 
interest  and  effort  and  by  a  larger  fund  of  relevant  past 


(After  Thorndike,  16,  p.  103) 



Gain  of  older  group    vy  

/->    '         r                                          ^  *^ 

Gain  ot  younger  group 


i.  Drawing  lines  of  given 
length,  blindfolded 
2.  Writing  with  wrong  hand 
3.  Transcribing  words  in  code 
4.  Memorizing  code  used  in  3 
5.  Learning  Esperanto 
6.  Memorizing  paired  num- 
bers and  syllables 
7.  University  studies 



over  I  oo  (estimated) 

Prisoners  of 

I.  Number-letter  substitu- 
tion from  key 
2.  Elementary  school  studies 
3.  Special  addition  practice 


High  School 

I.  High  school  studies:  Eng- 
lish, algebra,  civics,  etc. 



I.  Typewriting 
2.  Stenography 

Approximately  95  * 


*  Older  group  was  30  or  over,  rather  than  35  or  over. 


experience.  Thorndike  gives  many  examples  of  this  fact, 
in  addition  to  his  main  findings.  Thus  the  group  of  uni- 
versity students  progressed  over  twice  as  fast  in  learning 
Esperanto1  as  9-  to  iS-year-old  pupils  in  a  good  private 
school,  although  the  former  had  spent  less  than  half  as 
much  time  as  the  latter  in  studying  the  language.  Sim- 
ilarly, the  adult  prisoners  learned  elementary  school  sub- 
jects faster  than  average  schoolchildren. 

the  attempt  further  to  interpret  his  results,  Thorn- 
tlliat  two  of  the  tasks  employed  with  the 
gr?up?,  ^-?  ..drawing  lines  blindfolded  and  inqi- 
.pf  a  code,  measured  "sheer  mo.difiabiUty, 

or  learning  ability,  more  directly  than  the  other  t&sks, 
since-thfi^jK^S  .relatively  independent  of  past  experience. 
Drawing  chiefly  from  the  findings  on  these  two  tests, 
he  ventured  the  estimate 
that  there  is  a  decline  in 
"  sheer  modifiability"  of  w 
approximately  15%  be-  p 
tween  the  ages  of  22  and 

42,  a  negligible  loss  of  less      °        w       20       so       40       so 

,  tS  TT        i  Chronological  Age 

than  i%  a  year.    He  also      p  ^  ^  A 

J      .  .  FIG.  32.    ESTIMATED  CURVE  OP  AGE 

presents,  on  this  basis,  an    CHANGES  IN  LEARNING  ABILITY.   (After 

estimated    curve    of    age   Thorndike>  l6<  P-  ^ 

changes  in  "sheer  modifiability"  from  childhood  to  mid- 

dle maturity.   This  curve  is  reproduced  in  Figure  32. 

The  study  of  adult  learning  was  extended  into  the 
higher  age  levels  by  Ruch  (12).  Three  groups  of  subjects 
were  employed,  40  in  each  of  the  age  groups  12  to  17, 
34  to  59,  and  60  to  82.  The  tasks  to  be  learned  included 
two  motor  and  three  verbal.  Of  the  motor  tests,  one 
involved  a  simple  visuo-motor  coordination  in  direct  vi- 

1  One  of  the  proposed  "universal  languages,"  characterized  by  a  very  system- 
atic and  logical  structure  in  contrast  to  current  languages. 



sion; 1  the  other  consisted  of  the  same  task  performed  in 
mirror  vision,  thus  involving  the  establishment  of  re- 
actions which  were  opposed  to  the  subject's  past  experi- 
ence. All  of  the  verbal  tasks  involved  the  learning  of 
paired  associates.  The  first  consisted  of  word  pairs  having 
some  meaningful  connection,  such  as  nest-owl,  soft-chair. 
The  second  was  made  up  of  "nonsense"  material,  such 
as  A  X  M  =  5,  N  X  M  =  C,  and  was  thus  relatively- 
neutral  in  regard  to  past  association.  The  third  necessi- 
tated the  breaking  down  of  old  associations,  the  material 
consisting  of  false  multiplications,  such  as  5  X  4  =  2> 

3  X  i  «  i. 

In  Table  X  will  be  found  the  averages  and  standard 
deviations  of  the  three  groups  on  each  test.  Without  a 
single  exception,  the  averages  show  a  progressive  decline 
from  the  young  to  the  middle  and  from  the  middle  to 
the  old  group.  The  age  differences  are  largest,  however, 
nTthose  tests  whicli  are  hindered, by  ordinary  past  experi- 
ence and  smallest  in  those  which  are  aided  by  such 


(After  Ruch,  12,  p.  277) 



Standard  Deviations 







Motor  learning: 

I.  Direct 







2.  Mirror 







Paired  associates: 

i.  Meaningful 







2.  Nonsense 







3.  False  multipli- 


1  06.  1 






*A  modified  form  of  the  Koerth  pursuit  rotor,  in  which  the  subject  must 
follow  a  rapidly  and  irregularly  moving  object  with  a  stylus. 


Ruch  demonstrates  this  by  comparing  the  critical  ratios 1 
of  the  differences  for  each  test.  These  are  shown  below 
for  the  differences  between  young  and  old  groups,  middle 
and  old,  and  young  and  middle,  respectively. 

Y-0  M-O  Y-M 

Motor:  Direct  vision                    6.1  5.4                    0.9 

Motor:  Mirror  vision                   8,5  6.4                    0.6 

Meaningful  associates                  5.4  2.3                     3.9 

Nonsense  associates                      7.2  4.0                    2.9 

False  multiplications                  n.6  4.7                    5.9 

In  both  the  Y-O  and  M-O  comparisons,  the  results  appear 
to  be  very  consistent.  Mirror  vision  learning  shows  greater 
age  decline  than  direct  vision  learning;  within  the  paired 
associates,  there  is  a  progressive  rise  in  difference  as  we 
go  from  the  meaningful  associations,  through  the  neu- 
tral nonsense  material,  to  the  interference  material  of 
the  third  test.  The  comparison  of  young  and  middle 
groups  yields  somewhat  less  consistent  results,  probably 
because  these  differences  are  all  small  and  not  very 

An  examination,  qf,  the  standard  deviations  in  Table  X 
reveals"  a  rather  consistent  rise  in  variability  with  age,  in 

XMJ-^BJ^      ,  papam*"'"  »•**«'*'»*'  «"-•  *«*'  .t.u,.,w.),, ,,      ,  ,         ,,i,,,  „  ,,  ,         ,          ,   ,  ,  '  * 

spite"  of  the  equally  consistent ,  .drop  ia  averages.  The 
standard  deviations  of  the  oldest  group  are  all  larger 
than  those  of  the  youngest  with  only  one  exception,  viz., 
the  mirror  vision  test.  In  the  case  of  the  middle  group, 
all  standard  deviations  are  higher  than  those  of  the 
youngest,  with  no  exceptions.  These  findings  are  in 
agreement  with  the  analysis  of  age  and  variability  given 

1  Critical  ratio  =  >  a  common  statistical  device  for  estimating  the 


"reliability"  of  a  difference.  When  the  critical  ratio  is  equal  to  3  or  more,  it 
indicates  that  the  obtained  difference  could  not  have  resulted  from  chance  errors 
of  sampling,  and  such  a  difference  is  therefore  regarded  as  reliable. 



^£jnay  jaow  attempt  to  synthesize  the  findings  of 
..these  various  investigations  and  to  evaluate  them  in  the 
light  of  the  studies  on  training  discussed  in  the  preceding 
.chapter*  If  we  think  of  all  mental  development  in  terms 
of  learning,  the  diverse  findings  both  on  the  upper  limit 
of  mental  growth  and  on  the  decline  of  ability  can  be 
fitted  into  an  intelligible  pattern.  It  might  be  objected 
that  the  learning  curve  shows  no  decline,  whereas  age 
curves  do.  This  apparent  inconsistency  results,  however, 
from  an  incomplete  statement  of  the  situation.  The 
problem  will  be  considerably  clarified  if  we  speak  of  age 
changes  in  specific  tasks,  as  we  do  in  the  case  of  learning, 
rather  than  discussing  mental  development  in  general. 
It  is  quite  true  that  the  cumulative  effects  of  learning  in 
everyday  life  will  increase  proficiency  indefinitely  in 
certain  tasks,  but  such  learning  will  just  as  surely  inter- 
fere with  the  performance  of  other  tasks.  If  the  general 
effect  of  any  specific  act  of  learning  upon  all  of  the  indi- 
vidual's behavior  is  considered,  it  becomes  apparent  that 
learning  may  cause  a  decline  as  well  as  a  rise  in  achieve- 

Thjo.dediiie^^^  psychological  tests 

with  age  is  no  longer-surprising  when  .;we,xealize  the  re- 
semblance of  all  such  .tests. to.  school  work.  We  should 
therefore  expect  that  the  longer  the  individual  has  been 
out  of  school,  the  more  chance  he  has  had  to  forget  what 
he  learned  as  a  child,  through  interference  from  other 

Although  in  his  everyday  life  the  adult  may  be  employ- 
ing much  that  he  learned  in  school,  he  is  at  the  same  time 
losing  many  school  habits,  such  as  working  with  a  specific 
time  limit,  following  directions  literally  although  he  may 


see  little  sense  in  them,  and  especially  working  with  mate- 
rials which  may  be  meaningless  and  of  no  apparent  use 
to  him.  When  a  schoolchild  is  confronted  with  a  psycho- 
logical test,  the  novelty,  strangeness,  and  apparent  pur- 
poselessness  of  many  of  the  things  he  is  asked  to  do  will 
not  disturb  him  unduly,  since  at  that  age  he  is  still  doing 
many  things  for  which  he  can  see  no  immediate  value. 
Such  tasks  are  accepted  by  the  child  as  part  of  his  every- 
day work.  Not  so  with  the  adult.  The  older  he  grows, 
the  more  he  concentrates  only  on  those  activities  which  are 
either  of  practical  significance  or  directly  pleasurable  to 
him.  The  reaction  of  many  adults  to  intelligence  tests, 
as  contrasted  to  that  of  schoolchildren,  illustrates  this 
difference.  To  most  adults,  such  a  test  is  either  foolish  or 
entertaining.  The  adult  is  far  more  sensitive  to  the  appar- 
ent impracticality  of  the  situation  than  the  child  who  is 
used  to  taking  tests  which  to  him  are  almost  equally 

That  adult  ability  does  not  decline  in  all  tasks  is  demon- 
strated By  the  obvious  Improvement  in  functions  related 
to  the  individual's  daily  work.  The  achievements  of 
many  people  progress  along  a  continuously  rising  line 
throughout  life. 

I^or  can  a  distinction  be  legitimately  made  between  the 
extent  of  a  person's  abilities,  which  increases  con^ai^tly 
wij:h  age,  and  the  level  or  difficulty  oi-ta^^w]^K,J^J>s 
capable  of  mastering.1  The  latter  is  definitely  dependent 
upon  the  former.  As  was  brought  out  in  the  discussion 
of  practice  and  variability,  thejcnore^^ 
Uar^dfcj&a^  A  problem  which 

is  commonly  regarded  as  difficult  and  which  can  be  solved 
by  only  a  few  individuals  is  one  which  involves  the  synthe- 
sis of  more  numerous  and  varied  types  of  learned  behavior, 

1  Such  a  distinction  has  been  suggested  by  Thoradike  (15), 


We  should  say,  for  example,  that  the  derivation  of  a 
formula  which  requires  a  knowledge  of  arithmetic,  alge- 
bra, trigonometry,  and  calculus  is  more  difficult  than  one 
which  can  be  derived  simply  by  the  application  of  princi- 
ples of  arithmetic  and  algebra.  If  we  define  the  difficulty 
of  a  task  objectively  in  terms  of  the  number  of  people  who 
can  perform  it  correctly,  it  will  unquestionably  prove  to 
be  a  function  of  the  number  of  different  specific  abilities 
involved.  Even  if  a  more  subjective,  popular  definition 
of  difficulty  were  suggested,  it  would  doubtlessly  be  found 
to  hinge  upon  the  same  principle. 

Many  of  the  previously  reported  findings  are  clarified 
by  such  a  view  of  mental  development.  ,,,Xhu3,,the  slower 
decline  in  intelligence  test  performance  of  adults  who 
have  had  a  longer  period  of  schooling  is  to  be  expected 
when  we  recall  that  such  education  is,  relevant  to  the 
tasks  on  the  tests  themselves.  The  formal  education 
which  is  characteristic  of  our  culture  is  of  a  verbal  and 
abstract  sort  and  it  is  from  these  fields  that  intelligence 
tests  draw  predominantly  for  their  material.  All  groups 
have  been  constantly  subjected  to  training,  but  of  a  dif- 
ferent sort.  In  a  test  of  manual  operations,  for  example, 
we  should  expect  a  more  rapid  decline  with  age  among 
the  " better  educated"  professional  classes  than  among 
certain  groups  of  factory  workers. 

JForj:he ;jsame  reason,  the  limit  of  intellectual  improve- 
ment, as  measured by  conyjioa .  .iirt,dUigCTP.e..test3, .  .will  be 
reached  later  by  those  groujgsj^hich  continue  their  formal 
schooling  to  a  later  age.  This  has  been  repeatedly  demon- 
strated in  studies  on  the  "point  of  cessation"  of  mental 
growth.  Corroboration  of  the  proposed  interpretation  of 
age  changes  in  mental  traits  can  also  be  found  in  the 
experiments  on  adult  learning.  Thusjfjv^^ 
that  Thorndike  found  a  difference  in  rate  of  decline  be- 


tween  the  .more,  "meaningless"  and  thejDooze,,  " 

ful"  and  useful  tasks  j  and  Ruch  found  a,  similar  difference 

between  those  tasks  which  were  aided  and  those  which 

were  hindered  by  the  common  training  furnished  in  ©ur 


Finally,  mention  should  be  made  of  a  possible  physio- 
logically determined  decline  in  mental  activity  with  age, 
apjLCLJrom.cliajiges  correlated  .with  beaming.  The  effect 
of  the  deterioration  of  necessary  structures  doubtlessly 
plays  a  part  in  the  marked  and  sharp  decline  in  all  psycho- 
logical functions  which  frequently  characterizes  very  late 
old  age  or  senescence.1  Such  obvious  handicaps  as  fall- 
ing vision  and  hearing,  and  muscular  and  neural  deteriora- 
tion can  hardly  fail  to  affect  all  the  individual's  activities. 
These  changes,  however,  do  not  set  in  to  an  appreciable*ex- 
tent  until  very  late  in  life  and  consequently,  cannot  very 
plausibly,  be.-o£foed  as  -an  explanation  ^ofthe-deeMfie  in 
mental  test  performance  during  earlier  -matority.  There 
are  cases,  furthermore3  in  which  serious  structural  handi- 
caps during  old  age  have  been  compensated  to  a  remark- 
able degree  by  interest,  effort,  and  the  advantages  of  past 
experience.  The  wide  individual  differences  found  at  differ- 
ent ages  also  bespeak  the  potency  of  specific  environmental 
circumstances  rather  than  physiological  conditions  charac- 
teristic of  a  given  life  period. 

It  would  seem  that  the  physical  handicaps  of  senescence 
are  of  the  same  nature  as  the  physical  inadequacies  of 
the  immature  child;  they  set  the  upper  limits  of  behavioral 
development  at  a  given  chronological  period,  but  they 
do  not  determine  the  degree  to  which  such  limits  will  be 
approximated.  It  seems,  also,  that  these  physically  set 

1  For  a  discussion  of  the  physiology  of  old  age,  and  for  descriptive  accounts 
of  behavior  in  senescence,  the  reader  is  referred  to  Child  (2),  Hall  (6),  and 
Hollingworth  (7). 


limits  are  always  much  higher  than  is  commonly  sus- 
pected, since  training  and  stimulating  conditions  can  at 
all  ages  accomplish  surprising  results. 


1.  Baldwin,  B.  T.,  and  Steelier,  L.  I.   "Mental  Growth  Curves 
of  Normal  and  Superior  Children/'  Iowa  Univ.  Stud.  Child 
Welfare,  1922,  2,  No.  I.  Pp.  61. 

2.  Child,  C.  M.  Senescence  and  Rejuvenescence.   Chicago :  Univ. 
Chicago  Press,  1915.  Pp.  481. 

3.  Courtis,  S.  A.    "Maturation  Units  for  the  Measurement  of 
Growth,"  Sch.  and  Soc,,  1929,  30,  683-690. 

4.  .   Growth   and  Development   in   Children    (repr.    from 

Advances  in  Health  Educ,}.    N.  Y.:  Amer.  Child  Health 
Assoc.,  1934.    Pp.  180-204. 

5.  Freeman,  F.  N.    Mental  Tests.    N.  Y.:  Houghton  Mifflin, 
1926.   Pp.  503. 

6.  Hall,  G.  S.  Senescence.  N.  Y.:  Appleton,  1923.  Pp.  517. 

7.  Hollingworth,  H.  L.    Mental  Growth  and  Decline.    N.  Y.: 
Appleton,  1927.  Pp.  396. 

8.  Jones,  H.  E.,  and  Conrad,  H.  S.  "The  Growth  and  Decline 
of  Intelligence,"  Genet.  PsychoL  Mon.,  1933,  13,  223-298. 

9.  Miles,  C.  C.,  and  Miles,  W.  R.    "The  Correlation  of  In- 
telligence Scores  and  Chronological  Age  from  Early  to  Late 
Maturity,"  Amer.  J.  PsychoL,  1932,  44,  44-78. 

10.  Miles,  W.  R.     "Age  and  Human  Ability,"  PsychoL  Rev., 
I933>  4Q>  99-123. 

11.  Ruch,  F.  M.    "Adult  Learning,"  PsychoL  Bull.,  1933,  30, 

12.  .     "The  Differentiative  Effects  of  Age  upon  Human 

Learning,"  /.  Gen.  Psychol.,  1934,  n,  261-286. 

13.  Teagarden,  F.  M.    A  Study  of  the  Upper  Limits  of  the  De- 
velopment of  Intelligence.    N.  Y.:  Teachers  College,  Colum- 
bia Univ.,  Bur.  Pub.,  1924.  Pp.  112. 

14.  Thorndike,  E.  L.     "On  the  Improvement  in  Intelligence 
Scores  from  Thirteen  to  Nineteen,"  /.  Educ.  PsychoL,  1926, 
*7,  73-76. 


15.  Thorndike,  E.  L.,  and  staff.   The  Measurement  of  Intelligence. 
N.  Y.:  Teachers  College,  Columbia  Univ.,  Bur.  Pub.,  1926. 
Pp.  616. 

1 6.  Thorndike,  E.  L.,  et  aL  Adult  Learning.  N.  Y. :  Macmillan, 
1928.   Pp.  335. 

17.  Thurstone,  L.  L.,  and  Ackerson,  L.    "The  Mental  Growth 
Curve  for  the  Binet  Tests,"  /.  Educ.  Psychol.,    1929,  20, 

1 8.  Weisenberg,  T.,  Roe,  A.,  and  McBride,  K.  E.    Adult  In- 
telligence. N.  Y.:  Commonwealth  Fund,  1936.  Pp.  155. 

19.  Willoughby,  R.  R.     "Family  Similarities  in  Mental  Test 
Abilities,"  Genet.  Psychol.  Mon.,  1927,  2,  235-277. 



In  the  discussion  of  age  changes  in  mental  traits  during 
childhood  as  well  as  senescence,  mention  has  been  made 
of  certain  upper  limits  of  development  set  by  physical 
condition.  As  was  repeatedly  pointed  out,  such  struc- 
turally imposed  limitations  are  less  important  than  is 
commonly  supposed,  since  the  individual  rarely  attains 
the  degree  of  development  set  by  his  physical  capacity  at 
any  age.  We  majr  still  inquire,  however,,  whether  physical 
differences  "among  individuals  exert  an ,  appreciable  influ- 
ence upon  their  mental  differences.  It  is  apparent  that 
extreme  sensory  defects,  for  example,  can  so  profoundly 
handicap  the  individual  that  even  special  training  will 
not  bring  him  up  to  a  normal  level  of  performance.  Simi- 
larly, other  parts  of  the  reacting  organism  may,  by  their 
deficiency  or  superior  condition,  play  a  part  in  the  de- 
velopment of  mental  traits.  It  should  be  kept  clearly  in 
mind,  however,  that  structural  characteristics,  insofar  as 
they  are  shown  to  influence  mental  development,  can  be 
regarded  only  as  necessary  conditions;  thepresence^pf  a 
rqguired  physical  f actor  Jdoes_ not  jinj^ 
havioral  development,  ^ut^  ^mply ,. m^kea J&e  «M^I JE28"" 

The  question  of  the  relationship  between  mental  and 
physical  characteristics  has  long  proved  a  fascinating 
one  to  man.  The  field  has  abounded  with  fanciful  specula- 
tions and  vain  hopes  that  a  person's  character  or  intelli- 
gence could  be  "read"  from  physical  signs.  Recently, 



the  problem  has  been  approached  more  empirically  and 
objectively  and  a  sizable  body  of  data  has  been  accumu- 
lated on  the  relationship  between  various  structural, 
physiological,  or  sensory  conditions  and  psychological 
traits.  There  are  a  number  of  well-known  types  of  mental 
and  emotional  disorders  which  are  directly  traceable  to 
the  effects  of  drugs  or  infection  on  the  nervous  system, 
the  deterioration  of  tissues,  and  the  like.  ^  Thus  general 
paresis  results  from  syphilitic  infection,  delirium  tremens 
from  excessive  habitual  use  of  alcohol,  cretinism  from  a 
deficiency  of  the  thyroid  gland,  encephalitis  lethargica  l 
from  a  bacterium  which  is  believed  to  enter  through  the 
nasal  passages.  Many  other  similar  conditions  could  be 

In  the  present  discussion,  however,  we  shall  deal  only 
wjQp^5g-^j5jjgg^"g;'egrees  of  variation  commonly  found 
^^j^.^—^.^1  range  Q£  individuals.  We  are  not  con- 
ceffiet^witK'"'in6re  extreme  and  pathological  conditions, 
since  they  are  relatively  rare  and  their  effects  upon  be- 
havior are  more  obvious.  Tli^j^^r^^^uig^£^  the 
relationship  of  behavior. traits, to  craju&i, and „ facial, Char- 
acteristics, ,  tady  build,  ptiysiolpgical  qonditio3a%  a&d  sen- 
sory defects  will  be  surveyed  briefly  and  illustrative 
investigations  reported.  Some  further  data  on  anatomical 
characteristics  with  special  reference  to  theories  of  con- 
stitutional type  will  be  presented  in  the  following  chapter. 


Popular  interest  in  the  size  and  shape  of  the  skull  was 
considerably  stimulated  by  the  pseudo-science  of  phre- 
nology,2 initiated  by  Gall  in  the  eighteenth  century. 

1  Often  popularly  called  "sleeping  sickness,"  although  not  identical  with 
African  sleeping  sickness. 

2  For  a  discussion  from  the  viewpoint  of  a  practicing  phrenologist,   see 
Fowler  (4). 


Phrenology  was  based  on  a  false  notion  of  the  functions 
of  the  various  parts  of  the  cerebral  cortex.  Thus  the 
phrenologists  claimed  that  each  part  of  the  brain  con- 
trolled a  particular  intellectual  or  moral  function  such 
as  mechanical  ingenuity,  veneration,  domestic  impulses, 
and  other  equally  complex  and  vaguely  defined  activities. 
They  asserted  further  that  the  over-  or  underdevelopment 
of  such  characteristics  could  be  detected  by  examining 
the  protrusions  on  the  skull.  The  location  of  a  particu- 
lar "bump"  was  taken  to  mean  that  the  function  whose 
cortical  area  was  supposed  to  be  beneath  it  was  over- 
developed in  the  given  individual. 

It  would  seem  unnecessary  to  refute  such  an  obviously 
untenable  doctrine  were  it  not  for  its  still  widespread 
popularity  among  the  general  public  and  its  lucrative 
current  practice  by  quack  "vocational  counsellors"  and 
similar  soothsayers  of  modern  times.  In  the  first  place, 
phrenology  is  founded  on  the  erroneous  assumption  that 
there  is  a  perfect  correspondence  between  the  shape  of 
the  skull  and  that  of  the  brain;  this  is  hardly  to  be  ex- 
pected in  view  of  the  several  layers  of  membrane  and 
the  cerebro-spinal  fluid  which  intervenes  between  the  two. 
Size,  furthermore,  cannot  be  regarded  as  a  sufficient  index 
of  degree  of  development  in  the  nervous  system;  the 
latter  seems  to  depend  more  closely  upon  the  complexity 
of  interrelation  of  the  microscopical  nerve  cells.  Experi- 
ments on  cortical  localization  have  demonstrated  a  con- 
nection between  certain  muscle  groups  or  sense  organs 
and  certain  brain  areas;  but  this  is  quite  unlike  the  phre- 
nologist's attempt  to  map  out  complex  personality  traits. 

Phrenologists  have  also  tried  to  show  that  cranial 
capacity  as  a  whole,  or  total  brain  size,  is  related  to  intelli- 
gence. As  in  all  their  data,  however,  they  draw  their  evi- 
dence from  selected  examples.  It  is  true  that  a  certain 


type  of  idiot,  the  microcephalic,  has  a  very  small  skull, 
but  there  are  also  idiots  with  normal  or  very  large  skulls. 
A  few  men  of  genius  may  be  found  with  very  large  brains,1 
but  some  are  likewise  found  with  small  brains.  The 
question  can  only  be  settled  by  accurate  measurement  of 
large  numbers  of  unselected  cases. 

Investigations  on  the  relationship  between  cranial  ca- 
^acity^nr^  Intellectual  achievement,  have  generally 
yielded  negative  results.  In  a  number  of  studies  in  which 
average  cranial  dimensions  of  bright  and  dull  groups  were 
compared,  the  data  are  ambiguous  and  difficult  to  inter- 
pret.2 The  differences  between  the  averages  are  always  ex- 
tremely small  and  occasionally  inconsistent  from  one  com- 
parison to  another.  In  many  cases  the  measures  taken  on  the 
living  skull  were  not  good  indices  of  brain  capacity.  The 
groups  employed  varied  widely  in  age  and,  especially  when 
children  are  included,  this  may  produce  a  spurious  relation- 
ship between  size  of  head  and  intelligence  since  the  older 
subjects  will  have  larger  heads  and  at  the  same  time  will  ob- 
tain higher  scores  on  intelligence  tests.  Finally,  the  esti- 
mates of  intelligence  were  frequently  crude  and  unreliable. 

The  first  well  controlled  study  on  cranial  measurement 
in  which  adequate  correlational  analysis  was  employed  is 
that  of  Pearson  (17).  Measures  of  head  length,  head 
breadth,  and  cephalic  index  3  were  obtained  on  three 

1 A  frequently  quoted  example  is  Daniel  Webster,  whose  head  circumference 
measured  24!  inches. 

2  For  a  survey  of  these  data,  see  Paterson  (16). 

„ ,,     ,    ,.    T    i  100  X  head  width    ,         .     - ,       ,  „  ,  f          , 

3  Cephalic  Index  = *  Length  of  head  is  measured  from,  the 

head  length 

space  between  the  eyebrows  to  the  farthest  projection  at  the  back  of  the  head; 

head  width,  or  breadth,  is  the  distance  from  left  to  right  sides,  measured  from 

the  points  of  maximal  protrusion  above  each  ear.   The  following  is  a  common. 

classification  of  cephalic  index: 

Dolichocephalic,  or  long-headed        C.I.  below  75 
Mesocephalic,  or  medium-headed      C.I.  between  75  and  80 
Brachycephalic,  or  broad-headed       C-I-  above  80 


groups,  including  1010  Cambridge  University  students, 
over  2200  12-year-old  schoolboys,  and  over  2100  12-year- 
old  schoolgirls.1  It  will  be  noted  that  age  was  held  con- 
stant among  the  children  by  selecting  only  12-year-olds. 
The  subjects  were  classified  into  intellectual  levels  on  the 
basis  of  teachers'  ratings  and  scholastic  records.  The 
correlations  between  intellectual  level  and  cephalic  index 
were  —  .06,  —  .04,  and  .07  among  the  university  students, 
schoolboys,  and  schoolgirls,  respectively.  For  length  of 
head,  the  correlations  in  these  three  groups  were  .11,  .14, 
and  .08,  and  for  breadth  of  head  .10,  .11,  and  .11.  These 
correlations  speak  for  themselves,  being  too  low  in  every 
case  to  indicate  any  appreciable  trend.  The  very  low  and 
inconsistent  correlations  with  cephalic  index  lend  no 
support  to  a  frequently  proposed  theory  that  the  "long- 
headed" individuals,  with  a  low  cephalic  index,  are  the 
more  intelligent,  nor  to  the  opposite  view,  also  occasionally 
voiced,  that  the  "broad-headed,"  with  high  cephalic 
index,  are  the  more  intelligent. 

More  recent  investigations  by  the  correlation  method 
have  in  general  substantiated  Pearson's  findings.  Mur- 
dock  and  Sullivan  (14)  report  a  correlation  of  .22  between 
head  diameter,  obtained  by  averaging  maximum  head 
width  and  maximum  head  length,  and  I.Q.2  on  about  596 
elementary  and  high  school  pupils.  By  the  use  of  I.Q.'s 
and  by  the  conversion  of  physical  measurements  into 
deviations  from  the  average  of  each  age-sex  group,  the 
influence  of  age  was  held  constant.  Sommerville  (27) 
found  correlations  of  .10,  .03,  and  .09  between  head 
length,  head  width,  and  head  height,  respectively,  of 
100  male  college  students  and  their  scores  on  the  Thorn- 
dike  Intelligence  Examination  for  High  School  Gradu- 

1  The  number  of  cases  differed  slightly  for  each  measure. 

2  Found  from  a  number  of  group  intelligence  tests. 


ates.  The  correlations  were  no  higher  between  intelligence 
test  scores  and  cranial  capacity  as  estimated  from  the 
three  given  head  dimensions.  Employing  one  standard 
formula  for  the  computation  of  cranial  capacity,  Sommer- 
,  ville  obtained  a  correlation  of  .11  with  intelligence  test 
score;  with  another  formula,  the  correlation  was  .10. 
Reid  and  Mulligan  (23)  found  a  correlation  of  .08  be- 
tween cranial  capacity  and  scholastic  achievement  on 
449  male  medical  students  in  Scotland.  Cranial  capacity 
was  calculated  by  taking  the  product  of  head  length, 
breadth,  and  height,  with  allowance  for  thickness  of  dif- 
ferent parts  of  the  cranium.  Scholastic  achievement  was 
determined  by  performance  on  standardized  examinations 
in  three  courses  which  were  taken  by  all  the  students. 

It  seems  quite  definitely  established,  then,  that  no 
appreciable  relationship  exists  between  intellectual  per- 
fofmance  and  either  absolute,  head  , size,  or,,  head,  shape;, as 
determined  by  the  ceghalic  index.  Some  dissenting  voices 
are  still  heard,  advocating  the  use  of  cranial  measurement 
in  the  diagnosis  of  intellectual  defect,1  but  their  evidence 
is  very  ambiguous  and  their  arguments  are  untenable 
and  inconsistent.  Further  data  on  more  detailed  cranial 
conformation  will  be  presented  in  the  following  section 
in  conjunction  with  facial  measurements. 


There  are  at  present  many  firmly  entrenched  beliefs 
regarding  the  "meaning"  of  various  facial  and  other  bod- 
ily characteristics.  The  high  forehead  as  an  index  of  in- 
tellectual talent,  the  shifty  gaze  to  denote  deceitfulness, 
the  firm  chin  and  square  jaw  of  determination,  the  taper- 
ing fingers  of  the  artist,  and  a  host  of  other  traditional 

1  Cf.,  e.g.,  Porteus  (21). 


associations  which  the  reader  can  easily  name  have  found 
their  way  not  only  into  poetry  and  fiction  but  into  the 
snap  judgments  and  " hunches"  of  everyday  life.  We  also 
frequently  hear  of  alleged  personality  differences  between 
blondes  and  brunettes,  between  blue-eyed  and  brown-eyed 
persons,  or  between  those  with  a  "convex"  and  those 
with  a  "concave"  profile.  Most  of  these  beliefs  are  joined 
loosely  under  the  general  term  "physiognomy."  The  latter 
is  still  being  practiced  as  a  financially  profitable  "system" 
of  character  analysis. 

A  series  of  very  accurate  investigations  to  check  many 
of  the  assertions  of  physiognomy  were  conducted  under 
the  general  direction  of  Hull  (12).  The  relationship  be- 
tween convexity  of  profile  and  several  personality  traits, 
which  is  stressed  by  many  commercial  physiognomists,1 
was  studied  by  Evans  (3).  The  subjects  were  25  college 
women,  all  of  whom  were  members  of  the  same  sorority. 
Such  a  group  was  chosen  because  of  their  close  acquaint- 
ance with  each  other  and  their  resulting  ability  to  rate 
each  other  with  a  fair  degree  of  accuracy.  For  the  same 
reason,  all  individuals  who  had  not  been  members  long 
enough  to  be  well-known  were  excluded  from  the  study. 
Each  girl  ranked  the  remaining  24  in  six  personality  traits, 
including  optimism,  activity,  ambition,  will  power,  domi- 
nation, and  popularity.  The  average  or  consensus  rank 
of  all  24  judges  on  each  girl  was  computed  as  a  final  es- 
timate for  each  trait.  The  subjects  were  also  rated  in  a 
similar  way  for  degree  of  blondness.  A  specially  devised 
mechanical  instrument  was  employed  to  read  off  directly 
the  "angle  of  convexity"  of  the  profile.  In  order  not  to 
omit  any  possibilities,  convexity  was  measured  in  five 
different  ways,  such  as  whole  face,  upper  face  only,  con- 

1  Cf.,  e.g.,  Blackford,  K.,  and  Newcombe,  A.    The  Job,  the  Man,  the  Boss. 
N,  Y.:  Doubleday,  Page,  1919  (p.  154). 


vexity  without  including  the  nose,  and  so  on;  height  of 
forehead  was  also  measured. 

The  correlations  between  each  of  the  measures  of  con- 
vexity or  height  of  forehead  and  each  of  the  six  personality 
traits  were  low  and  often  inconsistent  with  expectation; 
thaTTs,  a  correlation  which  would  have  been  expected  to 
be  negative  on  the  basis  of  the  physiognomists'  claims  was 
positive,  and  vice  versa.  The  highest  correlations  were  a 
+.39  between  convexity  of  whole  face  with  nose  omitted 
and  " activity"  rank,  and  a  —.39  between  height  of  fore- 
head and  "will  power"  rank.  Even  these  correlations, 
however,  are  not  significant  in  view  of  the  small  number 
of  subjects,  and  they  could  have  resulted  from  chance 
errors  of  sampling.  The  correlations  for  blondness  range 
from  +.28  with  will  power  to  —.26  with  optimism.  These 
are  also  too  low  to  be  significant. 

A  further  point  to  bear  in  mind  in  evaluating  these  cor- 
relations is  that  the  existence  of  a  widespread  bias  among 
the  judges  regarding  the  association  of  facial  and  person- 
ality characteristics  might  in  itself  produce  a  correlation. 
Since  tests  were  not  available  for  the  traits  under  con- 
sideration, it  was  necessary  to  resort  to  associates'  judg- 
ments; but  this  procedure  is  inconclusive  when  prevalent 
popular  beliefs  are  present. 

Facial  and  cranial  measurements  were  combined  in  a 
study  by  Sherman  (25).  A  group  of  78  freshmen  in  an 
engineering  college  were  measured  by  means  of  a  specially 
designed  "radiometer."  A  total  of  15  distances  and  4  an- 
gles were  obtained  on  each  subject,  and  each  of  these 
measures  was  then  correlated  with  academic  grades  as  an 
index  of  scholastic  achievement.  The  correlations  with 
the  combined  grades  on  all  courses  ranged  from  —.26  to 
+  .34.  It  is  interesting  to  note  that  height  of  forehead 
correlated  —.15  with  academic  grades.  This  corroborates 


the  low  negative  correlations  found  by  Evans  between 
height  of  forehead  and  several  personality  traits  which 
might  be  expected  to  manifest  themselves  in  school  work. 
If  such  a  tendency  were  established,  it  would  indicate  a 
reversal  of  the  popular  notion  of  a  "high-brow" ! 

Various  similar  studies  have  been  conducted  by  many 
investigators.  In  Hull's  laboratory,  an  investigation  was 
carried  out  to  determine  whether  there  is  any  relationship 
between  the  shape  of  the  hand  and  a  number  of  traits 
suggested  by  "chirognomists."  The  results  were  definitely 
negative.  Numerous  experiments  have  been  conducted  to 
discover  whether  it  is  possible  to  judge  intellectual  or 
emotional  traits  from  photographs,  as  ,we  should  expect 
if  these  traits  were  manifested  in  facial  characteristics. 
All  of  these  investigations,  although  revealing  many  in- 
teresting cases  of  agreement  among  judges  which  suggest 
widespread  popular  beliefs  or  a  conventionalized  facial 
symbolism,  showed  no  agreement  with  independent  crite- 
ria of  the  traits,  such  as  intelligence  test  scores  or  observa- 
tions of  behavior. 

It  should  be  pointed  out  in  conclusion  that  even  when 
significant  correlations  are  found  between  certain  facial 
or  cranial  characteristics  and  psychological  traits,  as  in 
the  case  of  a  few  of  Sherman's  measures,  the  correlations 
are  still  too  low  to  give  any  information  about  individuals. 
They  simply  indicate  a  general  trend  in  the  group  which 
may  result  from  a  few  extreme  cases,,  Insofar  as  the  cor- 
relation is  far  below  i.oo,  it  shows  that  there  are  many 
individual  exceptions  to  the  general  trend.  The  presence 
of  such  exceptions  or  reversals  of  relationship  proves  that 
whatever  influence  any  physical  factor  may  exert  upon 
behavioral  development  is  very  minor  and  can  easily  be 
obscured  by  other  more  potent  factors. 

A  further  fact  to  note  is r  thj^as^lon^as  a  certain  notion 


]sjvidejy, .prevalent, reg&cdJBg  the,  association  of,  a  giypn 
physical  characteristic  with  a  mental  or  emotional  trait, 
thiTTnay  in  itself  influence  the  individual's  develppmeut. 
If  a  person  is  commonly  mistrusted  by  his  associates  and 
is  not  given  any  responsibility,  it  is  difficult  for  him  to  be 
open  and  sincere;  if  a  child  is  regarded  as  dull  and  stupid, 
he  may  easily  come  to  believe  it  himself  and  act  accord- 
ingly. The  social  and  motivational  influence  of  a  wide- 
spread prejudice  cannot  be  ignored.  A  vicious  circle 
is  initiated  by  such  a  situation;  the  more  widespread 
the  prejudice,  the  more  effective  it  will  be  and  the  more 
evidence  can  therefore  be  found  which  seems  to  sup- 
port it. 

From  these  considerations  it  is,  apparent  that  any  rela- 
tionship which  may  exist  between  facial  characteristics 
and  psychological  traits  cannot  be  large.  Even  the  very 
low  degree  of  correlation  occasionally  found  is  far  from 
being  conclusively  established  because  of  many  uncon- 
trolled factors.  Should  a  slight  correspondence  be  proved 
between  certain  facial  or  cranial  conformations  and  behav- 
ior, it  has  beeiijsuggested  that  such  association  may  result 
from  a  common  dependence  of  both  types  of  characteris- 
ticTujpbn  some  underlying  condition.  T^e,, activity f  of  the 
en4ocrine  glands  offers  possibilities  for  such  a  Connection. 
In  certain  extreme  pathological  cases  as,  for  example, 
thyroid  deficiency,  the  resulting  condition  includes  char- 
acteristic physical  as  well  as  mental  symptoms.  It  is 
barely  possible  that  certain  facial  characteristics,  as  well 
as  emotional  or  intellectual  traits,  are  influenced  within 
their  normal  range  of  variation  by  over-  or  underac- 
tivity  of  some  endocrine  gland.  This,  of  course,  is  only 
speculation.  The  field  of  endocrinology  is  far  too  com- 
plex and  too  young. to  offer  any  clear-cut  answers  to  such 
a  query. 



Gross  bodily  dimensions,  proportion  of  trunk  and  limbs, 
height  in  relation  to  weight,  and  similar  structural  char- 
acteristics have  also  been  suggested  as  possible  indices  of 
intellectual  or  emotional  status.  Since  much  of  the  mate- 
rial in  this  field  has  been  collected  to  test  out  the  various 
"type  theories"  proposed  from  time  to  time,  the  discus- 
sion in  this  section  will  be  supplemented  in  the  following 
chapter.  Only  the  data  on  gross  size  and  absolute  measures 
will  be  treated  here,  the  material  on  relative  proportions 
and  body  type  being  reserved  for  Chapter  IX. 

Similarly,  we  are  not  concerned  with  gross  malforma- 
tions and  pathological  conditions.  Many  of  these  condi- 
tions, with  which  anyone  who  has  seen  circus  "freaks" 
is  familiar,  have  been  definitely  attributed  to  glandular 
disorders.  Thus  gigantism,  a  condition  in  which  the  in- 
dividual may  attain  a  height  of  seven  or  eight  feet,1  results 
from  oversecretion  of  a  pituitary  hormone;  dwarfism,  or 
stunted  growth  in  which  the  bodily  proportions  are  nor- 
mal, is  produced  by  insufficient  pituitary  secretion.  No 
definite  intellectual  defect  has  been  demonstrated  in  these 
cases.  Cretinism,  associated  with  an  underactive  thy- 
roid, is  characterized  by  abnormal  bodily  development 
and  proportions  as  well  as  by  intellectual  defect,  sluggish- 
ness, and  other  behavioral  disturbances.  If  we  exclude 
cases  which  manifest  obvious  glandular  dysfunctions  or 
other  pathological  conditions,  we  still  find  a  wide  range  in 
height  and  weight  within  the  general  population.  It  is 
into  the  relationships  of  these  variations  with  behavioral 
characteristics  that  we  now  wish  to  inquire. 

As  in  the  case  of  cranial  measurements,  interest  in  body 
build  has  long  been  manifested.   The  search  for  a  possible 

1The  "giant"  with  the  Ringling  Bros.-Banrum  &  Bailey  circus  is  reported 
to  be  8  feet,  6  J  inches  tall 


relation  between  body  dimensions  ^nd  intellect  probably 
received  a  strong  impetus  from  the  popular  view  that  the 
intellectually  gifted  were  deficient  in  other  respects,  and 
in  particular  that  such  individuals  were  weak,  puny,  and 
physically  inferior.  This  notion  of  compensation  was 
cherished  widely  because  of  its  consoling  character — it 
was  no  doubt  accepted  as  the  device  of  a  benevolent  na- 
ture to  "even  things  up."  In  the  effort  to  overthrow  theae 
unfounded  beliefs,  early  research  workers  swung  to  the 
opposite  extreme  and  asserted  that  the  intellectually 
ablest  were  also  the  physically  ablest,  and  that  a  close  cor- 
respondence exists  between  physique  and  meatal  ability. 

Galton  (5),  for  example,  suggested  that  the  number,. of 
physically  superior  individuals  among  his  groups  of  emi- 
nent" men  (cf.  Ch.  IV)  was  greater  than  in  the  general 
population.  Many  studies  on  large  groups  of  children 
have""  subsequently  appeared  which  relied  upon  the  com- 
parison of  averages  for  their  conclusions.1  Such  investiga- 
tions agree  in  finding  a  slightly  higher  average  height  and 
weight  among  the  intellectually  superior  groups  than 
among  the  normal,  and  slightly  higher  for  the  normal  than 
for  the  dull.  Intelligence  was  usually  estimated  quite 
crudely  from  school  status  or  teachers'  ratings.  The 
differences  in  averages  were  invariably  so  slight  and  the 
overlapping  of  groups  so  large  that  the  degree  of  correla- 
tion between  height  or  weight  and  intelligence  would 
necessarily  be  negligible. 

Investigations  on  the  physical  status  of  the  feebleminded 
or  the  intellectually  gifted  child  have  yielded  results  which 
are  equally  difficult  to  interpret.  When  averages  are 
compared,  the  feebleminded  appear  to  be  definitely  below 
the  norms  in  height  and  weight,  and  the  bright  children 
above  the  norms.  In  Terman's  extensive  investigation  (29) 

1  For  a  summary  of  this  literature,  see  Paterson  (16). 



on  gifted  children,1  a  slight  tendency  was  noted  for  the 
subjects  to  be  above  the  age  norms  for  American-born 
children  In  height  and  weight.  L.  S.  Hollingworth  (n) 
compared  the  heights  of  three  groups,  each  composed  of 
45  children  between  the  ages  of  9  and  1 1 .  In  the  "  superior" 
group  were  only  children  whose  I.Q.'s  were  above  135 
(median  LQ.  =  151),  in  the  "normal/'  those  with  LQ.'s 
between  90  and  no  (median  LQ.  =  100),  and  in  the  "in- 
ferior/' those  with  I.Q.'s  below  65  (median  LQ.  =43). 
The  subjects  in  the  three  groups  were  carefully  equated, 
each  child  in  the  one  group  being  "matched"  with  a  child 
in  the  other  two  groups  in  respect  to  age,  sex,  and  racial 
background,,  so  that  the  influence  of  these  factors  was 
ruled  out.  In  Table  XI  will  be  found  a  frequency  dis- 
tribution showing  the  number  of  children  in  each  group 
who  fell  within  successive  class-intervals  in  height,  as 
well  as  the  average  height  of  each  group. 




(After  L.  S.  Hollingworth,  n,  p.  80) 

in  Inches 


Group  A 
(Median  I.  Q.  =  151) 

Group  B 
(Median  LQ.  =  100) 

Group  C 
(Median  LQ.  = 
















(15),  in  her  comprehensive  survey  of  the 
characteristics  of  the  feebleminded,  obtained  measures  of 

1  Cf.  Ch.  XIII  for  fuller  report. 


height  and  weight  on   157  mental  defectives  in  special 
classes  and  in  various  institutions.     She^  found  the  same 

marked  overlapping, 
the  mentally  defective  children  exceeding  the 

median  of  normal  children  in  weight  and  45%  in  height.1 
Goddard  (7)  collected  extensive  data  on  the  height  and 
welgHt  of  about  n?ooo  mentally  defective  individuals, 
ranging  in  age  from  early  infancy  to  60  years,  in  19  Amer- 
ican institutions  for  the  feebleminded.  In  Figures  33  and 
34  are  reproduced  curves  showing  the  average  height  and 
weight  of  successive  age  groups  within  four  intellectual 
levels;  the  data  on  boys  are  given  in  Figure  33,  those  on 
girls  in  Figure  34.  IJijmlLJ^  the  curves  of  the 

four  intellectual  groups  are  practically  identical  at  the 
lo^,,  levels;  the  lower  average  iieight  and  weight  «  of 
the  mentally  deficient  groups  becomes  apparent  oaly»as 
adolescence  and  maturity  are  approadtied. 

Several  factors  may  enter  in  to  complicate  the  analysis 
of  institutional  data  on  the  feebleminded.  In  the  first 
place,  those  individuals  with  physical  as  well  as  intellectual 
defects  are  more  likely  to  be  committed  to  an  institution. 
The  feebleminded  person  who  is  physically  fit  or  superior 
is  less  likely  to  be  sent  to  an  institution  at  all  and  more 
likely  to  leave  the  institution  after  he  has  received  several 
years  of  training.  Such  individuals  will  have  a  greater 
chance  to  succeed  in  a  routine  occupation  requiring 
strength  and  a  good  physique,  with  a  minimum  of  thought 
and  planning.  The  operation  of  such  a  selective  factor 

r  °  "**^»*^^*.wi*»4*<^M*p-^'^  .  y'  'V^v^^'^'%''V7'-,^nr^wvrt 

might  explain  the  divergence  of  Goddard  s  height  and 
we^KiTcurves  with  age.  Since  only  ^ns^^^^^^^Qre 
tested,  tKe  inferiority  at  the  upper  ages  could  have  resulted 
from  the  fact  that  the  physically  strongest  and  ablest  had 

1  Complete  overlapping  would  have  been  indicated  if  50%  of  the  feeble** 
minded  group  had  exceeded  the  normal  median. 






*  54 


|  50 






I     I     !     I     i     I     8     i 









20       22        24 


12       14       16       18 

Age   in  Years 

BOYS  AT  SUCCESSIVE  AGES.    (After  Goddard,  7,  p.  228.) 




-i  130 

8        10 

22       24 


12       14       16       18 

Age  in  Years 

GIRLS  AT  SUCCESSIVE  AGES.  (After  Goddard,  7,  p.  229.) 


left  the  institution.  In  addition,  the  norms  in  terms  of 
which  these  groups  are  evaluated  may  not  be  comparable 
at  successive  ages.  As  such  norms  are  usually  established 
on  schoolchildren  because  of  their  greater  accessibility  for 
measurement,  the  norms  at  higher  ages  are  frequently 
derived  from  high  school  students,  a  distinctly  select 
group  in  respect  to  the  general  population.  Finally,  when 
fairly  low  grade  feebleminded  subjects  are  employed,  it  is 
very  likely  that  several  cases  presenting  special  conditions 
such  as  cretinism  are  included  and  this  would  serve  further 
to  complicate  the  issue. 

results  obtained  more  recently 

by  the  use  of  the  correlation  technique  within  "'normal" 
"  groups,  wilLbe,  reported  as  more  conclusive.  Murdock  and 
Sullivan  (14),  in  the  investigation  cited  above  x  found  a 
correlation  of  .16  between  I.Q.  and  weight  (N  =  595)  and 
one  of  .14  between  I.Q.  and  height  (N=  597).  In  an 
investigation  on  58  children  in  the  kindergarten  and  57  in 
the  fourth  grade,  Gates  (6)  reports  a  correlation  of  .06 
between  Stanford-Binet  mental  age  and  height,  and  .10 
between  Stanford-Binet  mental  age  and  weight.2  Pearson 
and  Moull  (18)  correlated  height  and  weight  with  esti- 
mates of  intelligence  in  groups  of  616  Jewish  boys  and 
580  Jewish  girls  living  in  London;  the  correlations  between 
intelligence  and  height  were  .12  for  boys  and  .11  for  girls, 
and  between  intelligence  and  weight,  .15  for  boys  and  .07 
for  girls.  None  of  these  values  indicates  a  significant 
degree  of  relationship^ 

Data  on  older  subjects  have  yielded,  the  same,  general 
results,  the  correlations  often  being  ,ev£JX  lower,  .since  tjie 
groups  were  mpre  select  and  covered  a  narrower  rangy. 

1  In  the  section  on  cranial  measurements. 

2  These  correlations  are  the  averages  of  four  separate  correlations  for  kinder- 
garten boys  and  girls,  and  fourth  grade  boys  and  girls. 


Brooks  (2),  employing  1118  subjects  between  the  ages  of 
13  and  20,  in  the  third  year  of  junior  high  school,  the 
normal  school,  or  the  freshman  year  of  college,  correlated 
measures  of  height  and  weight  with  performance  on  sev- 
eral standardized  group  intelligence  tests.  Since  correla- 
tions were  computed  separately  for  the  two  sexes  and  for 
several  age  groups,  the  subjects  were  classified  into  17 
groups  ranging  in  number  of  cases  from  16  to  139.  The 
height  correlations  ranged  from  —.09  to  +.26;  those  of 
weight  ranged  from  —.31  to  +.26.  Finally,  in  the  study 
by  Somrnerville  (27)  on  college  freshmen  described  above,1 
a  correlation  of  .16  was  found  between  intelligence  and 
standing  height,  .13  between  intelligence  and  sitting 
height,  and  .10  between  intelligence  and  weight.  The 
majority  of  these  correlations  are  positive  but  so  low  as  to 
indicate  little  or  no  appreciable  relationship  between  gen- 
eral bodily  size  and  intellectual  level  when  selective  fac- 
tors and  other  irrelevant  conditions  are  ruled  out. 


Attempts  have  also  been  made  to  investigate  the  effects 
upon  mental  efficiency  of  various  physiological  conditions, 
such  as  general  health,  malnutrition,  defective  breathing, 
and  focal  infection  from  diseased  tonsils,  dental  caries, 
and  other  sources.  QuaM.lte,  first  comprehensive  surveys 

Sta-*  "''  M»>t*i>.*  u,«*  .  •  <  fr^  <M,«m  •?***"»  *  ""••,  i'«ww/v,«  fwyjwi" 

of  the  relative  frequency  of  physical  defects  among,  diil- 

''-'-»  ^^.-^^i^^  '*  +>..  "^"  -'•«>-'  -rt  -www^'^r'^'W'  v>,#"  t-~mtiti9vr 

A  total  of  3304  schoolchildren  between  the  ages  of  10  and  14 
were  classified  into  dull,  normal,  and  bright  groups  on  the 
basis  of  school  progress.  The  percentage  of  children  in  each 
category  having  various  common  defects  is  shown  below  (i, 
p,  74).  The  percentage  having  any  defect  at  all,  as  well  as 
the  average  number  of  defects  per  child,  are  also  given. 

1  Section  on  cranial  measurements. 


Nature  of  Defect 





Enlarged  glands 




Defective  vision 




Defective  breathing 




Defective  teeth 




Hypertrophied  tonsils 








Other  defects 








Average  number  of  defects 

per  child 




With  only  one  exception  every  comparison  reveals  a 
consistent  decrease  in  frequency  of  defects  from  the  dull 
to  the  normal  and  from  the  normal  to  the  bright  group. 
The  one  exception  is  defective  vision  which  shows  the 
opposite  relationship  from  all  the  other  defects,  being 
most  frequent  among  the  bright  and  least  frequent  among 
the  dull  subjects.  The  greater  studiousness  of  the  academ- 
ically superior  child  may  be  more  conducive  to  eye-strain 
and  thus  account  for  this  inconsistency,  perhaps  the  most 
noteworthy  finding  of  this  investigation  is  the  extremely" 
low  degree  of  relationship  indicated  between  physical  de- 
fect and  scholastic  achievement. 

"  Little  or  no  relationship  has  been  found  between  mental 
traits  and  such  conditions.^  ^general  health  and  mal- 
nutrition. This  problem  has  been  very  extensively  in- 
vestigated in  England  by  Pearson  and  his  students,  with 
Consistently  negative  results.  Correiatieae^etwsen, .these 
general  physical  conditions,, and  mtij^  on 

schoolchildren  in  various  social  and  ecoiioiuic_kvels  ^  in- 
VS£!SMZ  Pf  9 -ec^  *°  k^ .'.veiy- close,  to-zero.1  In  this  country, 
Hoefer  and  Hardy  (10)  conducted  an  intensive  investiga- 
tion on  343  third  and  fourth  grade  schoolchildren  between 
the  ages  of  8  and  II.  All  the  children  were  American-born 

lCf.,  e.g.,  Pearson  and  Moull  (18). 


Whites;  none  was  feebleminded  and  none  physically  de- 
formed. Thus  the  influence  of  extraneous  factors  was 
considerably  reduced.  Yearly  measurements  were  made 
on  the  subjects  over  a  period  of  three  years.  At  these 
times,  each  child  was  examined  by  a  physician,  anthropo- 
metric  measurements  were  taken,  and  finally  the  Stanford- 
Binet  and  the  Stanford  Achievement  Test  in  school  sub- 
jects were  administered.  Below  will  be  found  the  major 
data  relevant  to  the  problem  under  consideration. 


s>         j  or     •    7  /-•     j-  -  Number  Initial  Monthly 

General  Physical  Condition:  QJ  Cases  JQ  ^    / 


1.  Good  145  104  1.30 

2.  Improved  131  105  1.25 

3.  Poorer  27  99  1. 12 

4.  Poor  or  fair  40  101  i.n 
Condition  of  Tonsils: 

1.  Tonsillectomy  before  initial  test  78  107  1.28 

2.  Tonsillectomy  after  initial  test  64  104  1.33 

3.  Normal  tonsils  53  103  1.20 

4.  Improved  tonsils  30  102  1. 20 

5.  Diseased  tonsils  118  102  1.19 

The  children  were  classified  into  four  groups  on  the  basis 
of  general  health  as  determined  by  the  physician.  This 
classification  was  made  in  terms  of  the  relative  condition 
at  the  beginning  and  end  of  the  experiment.  Thus  "good" 
physical  condition  means  "good"  both  at  initial  and  final 
examination;  "improved"  and  "poorer"  indicate  the  di- 
rection of  change;  and  "poor  or  fair"  refers  again  to  an 
unchanged  condition  on  both  occasions.  The  subjects 
were  also  divided  into  five  groups  on  the  basis  of  the  con- 
dition of  their  tonsils,  as  indicated.  In  each  case,  the 
average  initial  I.Q.  of  the  group  is  given,  as  well  as  the 
average  monthly  gain  in  mental  age  months  during  the 
experimental  period.  The  intellectual  differences  among 


the  groups  are  in  all  cases  too  small  to  be  of  much  sig- 
nificance, and  the  relative  status  of  the  physically  better 
and  poorer  groups  is  often  inconsistent  with  expectation. 

In  view  of  the  frequent  exaggerated  claims  made  for 
the  effect  of  diseased  tonsils  upon  intellectual  status,  a 
direct  and  well  controlled  investigation  of  this  specific 
problem  was  conducted  by  Rogers  (24).  The  subjects 
included  530  public  school  boys  between  the  ages  of  6 
and  14.  All  had  been  given  the  Stanford-Binet  as  a  part 
of  the  regular  school  routine.  On  the  basis  of  an  examina- 
tion by  the  school  nurse  or  physician,  the  children  were 
classified  into  two  groups,  the  one  composed  of  236  boys 
whose  tonsils  were  sufficiently  diseased  to  require  treat- 
ment, and  the  other  of  294  boys  whose  tonsils  were  either 
not  defective  or  so  slightly  defective  as  not  to  deserve 
treatment.  The  Average I.Q«'s  of  the  normal  and  defective 
groups  proved  tp  be  954. and  94.9,  respectively.  The  per- 
centage distribution  of  the  two  groups  is  given  in  Fig- 
ure 35.  The  practically  complete  overlapping  of  these 
groups  is  apparent  from  an  examination  of  the  distribu- 
tion curves. 

In  order  further  to  check  upon  any  possible  influence  of 
tonsillar  condition  upon  mental  development,  28  boys 
whose  tonsils  were  subsequently  removed  were  retested 
with  the  Stanford-Binet  after  a  six  months'  interval.  The 
gain  in  LQ.  made  by  this  group  was  compared  with  that 
of  a  control  group  of  28  boys  who  suffered  from  diseased 
tonsils  but  who  had  not  been  operated  upon.  The  operated 
group  made  an  average  gain  of  2.25  LQ.  points  as  com- 

Siw»,i«,*«*Au«-*l-  ,u,  ,,„,,,  «»'  '"  >   •     •  -.'»'      ,    ','>(,,  ,,<•?>,    ^,]~3li         ,  ,i       '    „  i,,,  ,   ,,*.  <-»V  >,, 

pared[  to  an  average  gain  of  3.28  in  the  control  group. 
Finally,  it  was  possible  to  test  21  subjects  after  an  interval 
of  from  10  to  17  months  following  the  operation.  The 
average  jgainjft.  l,Q..tff&&&ky ,  tH^,gmWR^as,, ,3*0 .points, 
while  a  control  group  of  21  cases^gained  6.2  points.  It  is 



•——  Normal 








70-       80-       90-      100- 
80        90        100      110 
Intelligence  Quotient 

AND  DISEASED  TONSILS.   (Data  from  Rogers,  24,  p.  29.) 

very  doubtful  whether  further  retests  after  a  longer  delay 
would  reveal  any  effect  of  the  tonsillectomy  on  intellectual 
development, ,  Jifejj^k^^ 

ity  of  th^defcctive  group  as  cempamd*  ta».the»|]iprmal 
group  at  the  outset  and  the  absewe  pi  4UyllIlpJpvement 
in  I.Q.  which  could  be  directly  attributed  to  the.,  removal 
oF^SIs^seH  tonsils  are  inutually,cpi:i:obQ,r^tiY^in ,  deuion- 
stfatmg  a  lack  of  relationship  between  ment^Ul  -lewl  and 
this  "particular  physical  condition. 


t|me  in  the  popular  press  i  and 

effects  oTHenf alvearies  (decayed,  t 

tions.    Although  there  is  relatively;  litjlp^i^forniation  on 

this  question,  what  there  is  seems  to  point  to  a  complete 


lack  of  relationship.  Pearson  and  Moull  (18),  for  example, 
report  a  correlation  l  of  .15  between  number  of  carious 
teeth  and  estimates  of  intelligence  on  a  group  of  617 
Jewish  boys  and  a  similar  correlation  of  .10  on  581  Jewish 
girls  living  in  London. 

A  rather  .suggestive  study  on  this  problem  was  con- 
ducted by  Kohnky  (13).  Two  fifth  grade  classes  in  the 
same  school,  closely  similar  in  all  respects,  were  selected 
for  the  experiment.  One  class,  consisting  of  38  children, 
served  as  the  control  group.  The  other,  comprising  40 
children,  received  all  possible  dental  attention  during  the 
school  year.  This  included  dental  treatment  whenever 
needed,  toothbrush  drills,  dinners  to  afford  opportunity 
to  teach  proper  mastication,  inspection  by  the  school 
nurse,  home  visits  by  a  social  worker,  etc.  A^  series  of 
sen^ori-mqtor  and  simple  mental  tests  were  administered 
to  both  groups  in  October  and  in  May.  The  differences  in 
gain  made  by  the  control  and  experimental  groups  were 
very  slight  and  inconsistent,  in  a  few  cases  the  control 
group  making  the  larger  gain. 

It  is  commonly  believed  that  hookworm  (Necator  amer- 
icanus)  infection  produces  mental  defect,  sluggishness,  and 
apathy.  Because  of  the  very  common  occurrence  of  this 
infection  among  schoolchildren  in  certain  parts  of  the 
country,  it  has  attracted  the  attention  of  educators. 
Several  studies  have  indicated  a  tendency  for  children 
with  hookworm  infection  to  be  duller  than  those  not  so 
afflicted.  An  investigation  by  Smillie  and  Spencer  (26) 
gives  typical  results.  A  group  of  118  children  in  grades 
3  to  7  of  three  rural  schools  in  a  hookworm  "area"  were 
given  the  Otis  Intelligence  Test.  Their  scores  were  trans- 

1  The  two  variables  correlated  in  this  study  were  so  arranged  that  a  positive 
correlation  would  indicate  more  carious  teeth  among  the  less  intelligent  in- 


muted  into  I.Q.'s  in  order  to  eliminate  the  disturbing 
factor  of  age.  On  the  basis  of  medical  tests,  it  was  possible 
also  to  determine  the  degree  of  hookworm  infestation  for 
each  child.  Below  are  given  the  average  I.Q.'s  of  children 
in  five  categories,  ranging  from  a  normal  group  who  showed 
no  trace  of  hookworm  infection  to  the  most  heavily  in- 
fected group  (from  26,  p.  319). 

Intensity  of  Infection  Number  of  Cases  Average  I.Q. 

Normal  17  90.2 

Very  light  (1-25)  l  40  88.3 

Light  (26-100)  27  86.4 

Moderate  (101-500)  23  84.1 

Heavy  (501-2000)  10  76.3 

Although  the  differences  in  averages  are  appreciable  if 
extreme  groups  are  compared,  the  overlapping  of  all 
groups  is  large.  When  individual  scores  rather  than  group 
averages  are  considered,  a  correlation  of  .30  2  is  obtained 
between  LQ.  and  degree  of  hookworm  infestation. 

This  correlation,  although  not  high,  indicates  a  some- 
what closer  degree  of  relationship  than  has  been  found 
between  mental  level  and  any  of  the  other  physiological 
conditions  so  far  discussed.  The^  analysis  of  results  ob- 
tained in  investigations  on  hookworm  suggests  the  ppera- 
tioEToFa  factor  whack  is  probably  pre^eat^  -^Ithougk  to  a 
lesser  extent,  in  all  studies  on  the  i-elatipnahipw between 
mental  and  physical  characteristics.  The  individuals  of 
inferior  physical  condition  in  general  tend  to  come,  from  a 
poorer  social  level:  their  environment  is  deficient  in  op- 

•T.    ,,       ..-.. "    "•><    *        ,,,,?.,  i   '      i    .,.  ,  -        ,.,..       „  f  »  ',.-^,,44r™>-"^*«'miA«rx/«f     A 

portunity  for  mental  developI^e^t.a 

conditions,  facilities  for  medical  at 

and  home  ?care, J'ancf  sp^fprtti.    This  is  particularly  well 

illustrated  by  hookworm,  a  condition  which  is  prevalent 

1  Estimated  number  of  hookworms. 

2  Computed  by  Paterson  (16,  p.  196),  from  the  data  of  Smillie  and  Spencer. 


among  individuals  of  low  social  status  and  flourishes  In 
very  poor  and  backward  rural  districts.  The  environmen- 
tal background  may  be  the  common  underlying  factor 
which  leads  both  to  the  physical  and  to  the  mental  con- 
ditions. This  could  in  itself  account  for  what  little  rela- 
tionship is  found  between  physical  condition  and  mental 


Finally,  we  may  inquire  into  the  effects  of  sensory  defi- 
ciencies upon  behavioral  development.  Visual  and  audi- 
tory defects  are  the  most  serious  sensory  handicaps  for 
man.  Since  our  culture  is  built  to  such  an  enormous 
jextmt  upXMX-  a  fpup3SJo5ij5i"J.aDigaage,  and  the  latter  is 
Acquired  chiefly.  through  the  eye  and  the  eary  the  signifi- 
cance of  deficiencies-,  in  these-  particular  sensory  fields  is 
apparent.  Sensory  limitations  have  a  much  more  direct 
bearing  upon  behavior  than  the  other  kinds  of  physical 
deficiency  discussed  above.  -Environmental  stimulation 
is  cut  off  by  blindness  or  deafness;  the  individual  so 
afflicted  is  psychologically  "isolated"  from  cultural  con- 
tacts in  the  same  sense  as  the  wolf  children  of  India  or 
Kaspar  Hauser,  described  in  Chapter  III.  We  should 
therefore  expect  a  fairly  pronounced  behavioral  deficiency 
to  be  associated  with  such  sensory  defects. 

Relatively  little  progress  has  been  made  in  the  psy- 
chological testing  of  the  blind.1  Hayes  (9)  has  adapted 
the  Stanford-Binet  for  testing  the  blind  and  has  also 
administered  several  group  intelligence  and  educational 
achievement  tests  to  blind  subjects, 
chMrgn_Lest  below  the  normsfor 

sion;  their  average  LQ.  has  been  estimated  a 
aT~compareH  to  100  for  children  in  general.     A  much 

'  and  Haine7(8)7  " 


greater  percentage  of  children  classified  as  distinctly 
feebleminded  are  also  found  among  the  blind.  There  seems 
to  be  no  evidence  for  the  popular  belief  that  the  blind 
have  a  finer  discrimination  than  the  normal  in  other 
senses,  such  as  hearing  or  touch.  The  remarkable  feats 
often  accomplished  by  blind  persons  through  the  use  of 
other  senses  result  from  a  more  efficient  use  of  sensory 
cues  rather  than  in  a  superiority  of  the  senses  themselves. 
Through  prolonged  training,  an  individual  may  acquire 
the  ability  to  respond  to  very  slight  cues  which  are  ordi- 
narily ignored.  Such  seems  to  be  the  case  among  the  blind. 
Minor  visual  defects  seem  to  have  little  or  no  effect  upon 
intellectual  development.  It  will  be  recalled  that  in  Ayres' 
survey,  for  example,  the  percentage  of  visual  defects  was 
slightly  larger  among  the  accelerated  and  slightly  smaller 
among  the  retarded  children  than  among  children  of 
normal  age-grade  location.  It  should  be  noted,  however, 
that  unlike  deficiencies  in  other  sensory  fields,  visual  de- 
fects are  frequently  corrected  by  the  use  of  lenses.  If  the- 
child  wears  glasses  from  a  very  early  age,  the  stimulational 
handicap  resulting  from  his  sensory  deficiency  will  be 
eliminated.  If  the  relation  between  sensory  defect  and 
intellectual  development  is  attributable  entirely  to  the 
psychological  "isolation"  produced  by  sensory  limita- 
tions, we  should  not  expect  to  find  any  mental  inferiority 
when  the  sensory  deficiency  has  been  compensated  by 
artificial  means. 


This  is  very  likely  owing  to  their  greater  difficulty  in  the 
acquisition  of  language.  The  language  ability  of  the  deaf 
in  general  has  been  found  to  be  very  deficient,  and  as  a 
result  their  whole  educational  progress  is  much  slower. 
Even  when  oral  instructions  are  eliminated,  the  deaf  can- 


not  be  tested  by  means  of  the  ordinary  intelligence  tests 
because  of  their  serious  language  handicap. 

Reamer  (22),  for  example,  reports  an  average  retarda- 
tion of  five  years  in  standardized  educational  tests  among 
deaf  children.  Pintner  (^^jjajaj^^ 

deaf,JpQund  aja  equally  .wide  discrepancy  between  the,  edu- 
cational achievement  of  deaf  and  hearing  children.  Com- 
paring the  performance  of  deaf  children  at  different  ages 
with  norms  on  hearing  children,  he  obtained  the  following: 

Age  P  Average  Score  ^  Educational  Quotient 

Expressed  as  an"  Educat-i  onal  Age 

iz  7  years~9  months  65 

13  8  years- 1  month  62 

14  8  years-9  months  63 

15  9  years  60 

Thus,  the  average  1 5-year-old  deaf  subject  made  a  score 
equivalent  to  that  of  the  average  g-year-old  hearing  child, 
and  so  on  for  the  other  ages.  The  educational  quotient 
given  in  the  third  column  is  to  be  interpreted  analogously 
to  the  I.Q. 

In  the  same  investigation  it  was  found  that  the  con- 
genitally  deaf  or  those  who  became  deaf  during  the  first 
few  years  of  life  manifested  greater  educational  retarda- 
tion than  those  who  acquired  deafness  later  in  life,  the 
most  marked  differences  being  observed  between  the 
three-year  and  the  five-year  groups.  Those  subjects  in 
whom  the  onset  of  deafness  was  postponed  until  the  fifth 
year  had  had  an  opportunity  to  attain  a  certain  level  of 
linguistic  development  before  the  necessary  stimulation 
was  cut  off. 

In  the  effort  to  rule  out  language  handicap,  specially 
devised  performance  and  non-language  tests  have  been 
employed  with  deaf  children.  In  one  very  comprehensive 
investigation  by  Pintner  (19),  a  total  0^4432  children,  12 


years  of  age  or  older,  were  tested  in  41  schools  for  the  deaf 
by  means  of  the  Binixi£^^  Test.  This  test 

involves  no  language  in  any  part,  the  directions  being 
given  by  demonstration  and  pantomime.  The  average 
score  made  by  children  in  each  age  group  was  transmuted 
into  a  mental  age  and  from  these,  I.Q.'s  were  computed. 
The  results  are  summarized  below: 

Age  Average  M.A.  Average  l.Q. 

12  10  years  83 

13  10  years-6  months  81 

14  ii  years  79 

15  12  years  80 

If  these  averages  be  compared  to  those  obtained  with 
educational  tests  (cf.  above),  the  greater  inferiority  on  the 
latter  will  be  apparent.  The^lisiination^of  even  written 
language  from,  the  test  situation  has  reduced  the  handicap 
of  the  deaf  child,  but  it  has  not  eliminated  it.  By  cutting 
off  an  important  source  of  environmental  stimulation, 
deafness  has  produced  a  more  widespread  effect  than  just 
a  linguistic  deficiency.  The  subject  is  backward  in  all 
intellectual  tasks,  whatever  their  nature. 

Mention  should  also  be  made  of  the  deaf-mute.  These 
individuals  present  a  very  striking  illustration  of  the  in- 
fluence of  environmental  stimulation  in  the  development 
of  a  function.  Never  having  heard  the  human  voice,  the 
deaf-mute  is  unable  to  speak  although  his  vocal  organs  may 
be  perfectly  normal.  The  presence  of  human  vocal  organs 
does  not  in  itself  lead  to  the  development  of  human  speech, 
any  more  than  any  other  structure  insures  the  presence  of 
the  function  ordinarily  associated  with  it.  Vocal  organs 
of  a  certain  type  are  a  necessary  but  not  a  sufficient  con- 
dition for  the  acquisition  of  speech.  That  the  deficiency  of 
the  deaf-mute  is  a  stimulational  one  is  demonstrated  by 
the  cases  in  which  such  persons  have  been  taught  to  speak 


through  the  use  of  other  cues.  The  remarkable  results 
obtained  in  a  few  such  cases,  of  which  the  blind  deaf-mute 
Helen  Keller  is  probably  the  most  famous  example,  again 
testify  to  the  importance  of  training  in  behavioral  de- 
velopment. When  even  such  intense  structural  handicaps 
as  complete  blindness  and  deafness  have  been  overcome 
by  intensive  training  to  such  an  extent  that  the  individual 
is  capable  of  normal  or  even  superior  intellectual  achieve- 
ment, the  factor  of  structural  limitation  seems  to  lose 
much  of  its  potency. 

In  regard  to  milder  hearing  deficiencies,  there  is  some 
evidence  to  suggest  a  possible  effect  upon  intellectual 
status.  Sterling  and  Bell  (28)  tested  1860  schoolchildren 
between  the  ages  of  8  and  17  with  an  audiometer.1  Having 
arbitrarily  defined  a  "significant  loss  of  hearing"  as  a  loss 
of  nine  or  more  units  from  the  average  reading  on  the 
audiometer  scale,  these  investigators  found  such  a  hearing 
deficiency  among  2.9%  of  the  retarded  children  and  only 
1.2%  of  those  in  the  normal  age-grade  location.  Similarly, 
when  the  children  were  divided  into  those  doing  excellent 
school  work,  those  whose  work  was  satisfactory,  and  those 
whose  work  was  unsatisfactory,  the  percentage  of  children 
with  hearing  deficiency  in  each  group  was  1.6,  1.7,  and 
2.2,  respectively.  I.Q.'s  were  available  on  585  of  the  total 
group  of  children.  When  these  subjects  were  classified  on 
the  basis  of  intelligence  test  performance,  the  following 
results  were  obtained : 

,.  ft  Percentage  of  Children  Showing 

a  Significant  Loss  of  Hearing 

Above  average  0.6 

Normal  1.6 

Below  average  3.7 

1 A  very  accurate  standardized  instrument  for  determining  the  weakest,  or 
least  intense,  audible  sound  for  each  ear. 


Although  very  small,  these  differences  are  interesting  in 
view  of  the  fact  that  the  hearing  deficiencies  under  con- 
sideration were  not  so  serious  as  to  produce  a  marked 
stirnulational  handicap. 


In  the  light  of  the  various  findings  on  the  relationship 
between jnental  and  physical  traits  which  have  been  sur- 
veyed in  the  preceding  sections,  we  may  inquire  into  the 
pO4§]bj£  natnt^oi  such  a  relationship.  We  are  confronted 
in  the  first  place  with  certain  abnormal  conditions  of  the 
organism  which  have  characteristic  physical  as  well  as 
behavioral  symptoms.  It  is  unwarranted,  however,  to 
generalize  from  the  association  found  in  such  pathological 
cases  to  a  possible  connection  among  individuals  in  gen- 
eral. To  take  an  obvious  and  extreme  illustration,  an 
individual  whose  legs  have  been  amputated  to  the  knee  is 
unable  to  dance;  we  cannot  conclude  from  this,  however, 
that  length  of  leg  is  correlated  with  dancing  ability  and 
that  those  persons  with  longer  legs  will  be  the  better 
dancers  within  a  group. 

^  ,d.emon- 

^  iittle^is 


ingjthe  qperationrti^  in  -behavioral 

development.  In  the  field  of  endocrinology,  for  example, 
much  remains  to  be  learned.  Too  often  what  has  been 
offered  as  a  stimulating  hypothesis  for  further  research 
has  been  interpreted  by  the  layman  as  an  established  fact. 
The  same  may  be  said  in  regard  to  nerve  physiology.  The 
field  abounds  in  speculation  and  the  experts  still  disagree. 
At  such  a  stage,  it  is  definitely  premature  to  venture  a 
systematic  analysis  of  behavior  in  terms  of  the  nervous 


If  we  turn  from  the  observation  of  pathological  cases 
and  speculations  on  the  physiological  mechanisms  under- 
lying behavior  to  data  collected  on  normal  groups,  we 
still  meet  difficulties.  Many  of  the  investigations  on  this 
problem  have  been  inadequately  controlled.  Through  the 
misinterpretation  of  statistical  techniques,  slight  general 
trends  in  groups  have  been  erroneously  attributed  to 
individual  cases.  It  will  be  recalled,  for  example,  that  the 
small  differences  in  group  averages,  which  were  regarded 
as  significant  by  many  early  workers,  actually  showed 
only  a  negligible  relationship  when  the  individual  scores 
were  correlated.  Tli^J^r^  groups 

was  pften_ig|XQ.lfid-  Age  differences  were  occasionally 
pTesent  within  the  groups,  thus  producing  a  spurious  con- 
nection between  certain  physical  characteristics  and  men- 
tal level. 


utmost  impprtance,Jn((,aJliy  i^ye?.t:ig^tiQA.Hpa  ^.J^l^lP^'- 
s^  of  jphysical  and  .mental  condition.  The  individual  of 
superior  social  level,  coming  from  a  better  home,  will  have 
richer  opportunities  for  intellectual  development  and  at 
the  same  time  receive  better  physical  care.  He  will  be 
brought  up  under  more  sanitary  conditions  and  will  have 
less  chance  of  contracting  disease  than  the  less  fortunate 
child  reared  in  a  city  slum  or  a  poor  rural  district.  Tins 
factor  is  probably  responsibly  .to  a  large  -extent  J^jyh at 
little  positive  correlation  has  been  found  between,  mental 
an^physical  traits. 

BearingTn  inind  the  fact  that  direct  investigation  on 
normal  groups  has  generally  yielded  very  low  positive 
correlations  between  mental  and  physical  characteristics 
and  that  this  correlation  would  be  still  further  reduced 
if  social  status  were  held  constant,  we  shall  examine  two 
possible  interpretations  of  such  a  relationship.  On  the 


one  hand  we  have  the  theory  voiced  by  Qalton  and  others 
in  the  attempt  to  refute  th.e  popular,  notion  of  "compen- 
sation." This  theory  interprets  the  positive  correlation 
between  mental  and  physical  traits  in  terms  of  some 
general  quality  of  the  organism  which  underlies,,  all  forms 
of  development.  It^ssumes_tha,t  "  good  things  go  to- 
gether" just  on  general  principle, 

tends  to  be  superior  in  others 


on  the,  direct 
pS&  This 

handicap  can  take  many  forms.  In  the  case  of  sensory 
deficiencies,  there  is  a  partial  stimulational  isolation  of 
the  individual.  Malnutrition,  poor  health,  and  other 
general  physiological  conditions  reduce  endurance,  increase 
fatiguability,  affect  muscular  development,  and  generally 
lower  the  efficiency  of  work.  These  conditions,  if  suf- 

^^  -a,  certain  ,  ,eztent.   Physical  defects 

or  discomforts  also  prove  a  powerful  distraction  and  would 
thus  make  it  more  difficult  for  the  child  to  concentrate 
on  his  school  work  or  other  tasks.  Finally,  certain  strik- 
ing facial,  cranial,  or  bodily  characteristics  which  have 
acquired  a  specific  significance  through  some  traditional 
belief  or  superstition  may  affect  the  individual's  subse- 
quent intellectual  and  emotional  development,  because  of 
the  social  attitudes  which  they  engender. 


on  the 

basis  of  direct  b^vjpiaLluadicap.    There  is  no  need  to. 
invoEe^TlJoysterious  underlying  quality  which  produces 
the  all-around  "good"  person  or  the  all-around  deficient 



1.  Ayres,  L.  P.     "The  Effect  of  Physical  Defects  on  School 
Progress,"  PsychoL  Clinic,  1909-10,  3,  71-77. 

2.  Brooks,  F.  D.    "The  Organization  of  Mental  and  Physical 
Traits  during  Adolescence,"  /.  Appl.  PsychoL,   1928,   12, 

3.  Evans,  A.  L.    The  Alleged  Relations  between  the  Face  and  the 
Character.     Unpub.  A.B.  Thesis,   Univ.  Wisconsin,    1921 
(also  reported  in  Hull,  12). 

4.  Fowler,  0.  S.    Human  Science  or  Phrenology.    Phila.:  Nat. 
Pub.  Co.,  1873.  Pp.  1211. 

5.  Galton,  F.    Hereditary  Genius.    London:  Macmillan,  1914. 
Pp-  379- 

6.  Gates,  A.  I.    "The  Nature  and  Educational  Significance  of 
Physical  Status  and  of  Mental,  Physiological,  Social,  and 
Emotional  Maturity,"  /.  Educ.  PsychoL,  1924,  15,  329-358. 

7.  Goddard,  H.  H.   "The  Height  and  Weight  of  Feebleminded 
Children  in  American  Institutions,"  /.  Nerv.  and  Mental 
Diseases,  1912,  39,  217-235. 

8.  Haines,  T.  H.      "Mental  Measurements   of  the  Blind," 
PsychoL  Hon.,  1916,  No.  I.  Pp.  86. 

9.  Hayes,  S.  P.     Terman's  Condensed  Guide  for  the  Stanford 
Revision  Adapted  for  the  Blind.   Watertown,  Mass.:  Perkins 
Inst.  for  the  Blind,  1930. 

10.  Hoefer,  C,  and  Hardy,  M.  C.   "The  Influence  of  Improve- 
ment in  Physical  Condition  on  Intelligence  and  Educational 
Achievement,"  2jth  Yearbook,  Nat.  Soc.  Stud.  Educ.,  1928, 
Parti,  371-387. 

11.  Hollingworth,  L.  S.   Gifted  Children,  Their  Nature  and  Nur- 
ture. N.  Y. :  Macmillan,  1926.  Pp.  374. 

12.  Hull,  C.  L.    Aptitude  Testing.    Yonkers-on-Hudson:  World 
Book  Co.,  1928.  Pp.  535. 

13.  Kohnky,  E.    "Preliminary  Study  of  the  Effect  of  Dental 
Treatment  upon  the  Physical  and  Mental  Efficiency  of 
Schoolchildren,"  /.  Educ.  PsychoL,  1913,  4,  571-578. 

14.  Murdock,  K.,  and  Sullivan,  L.  R.    "A  Contribution  to  the 


Study  of  Mental  and  Physical  Measurements  in  Normal 
Children/'  Amer.  Phys.  Educ.  Rev.,  1923,  28,  209-215; 
276-280;  328-330. 

15.  Nors worthy,  N.     "The  Psychology  of  Mentally  Deficient 
Children,"  Arch.  PsychoL,  1906,  No.  I.   Pp.  in. 

16.  Paterson,  D.  G.    Physique  and  Intellect.    N.  Y.:  Century, 
1930.   Pp.  304. 

17.  Pearson,   K.      "Relationship  of  Intelligence  to  Size   and 
Shape  of  the  Head  and  Other  Mental  and  Physical  Char- 
acters," Biom.,  1906,  5,  105-146. 

1 8.  Pearson,  K.,  and  Moull,  M.    "The  Problem  of  Alien  Im- 
migration into  Great  Britain,  Illustrated  by  an  Examination 
of  Russian  and   Polish  Jewish   Children,"  Part  II,  Ann. 
Eugenics ',  1925-26,  I,  56-127. 

19.  Pintner,  R.     "A  Mental  Survey  of  the  Deaf,"  /.  Educ. 
Psychol.,  1928,  19,  145-151. 

20.  .     Intelligence  Testing:  Methods  and  Results.     N.  Y.: 

Holt,  1931.  Pp.  555. 

21.  Porteus,  S.  D.,  and  Berry,  R.  J.  A.   "Intelligence  and  Social 
Valuation,  a  Practical  Method  for  the  Diagnosis  of  Mental 
Deficiency  and  Other  Forms  of  Social  Inefficiency,"  Fine- 
land  Training  Sch.  Res.  Pub.,  No.  20,  1920.   Pp.  100. 

22.  Reamer,  J.  C.    "Mental  and  Educational  Measurements  of 
the  Deaf,"  Psychol.  Man.,  1921,  29,  No.  3.  Pp.  130. 

23.  Reid,  R.  W.,  and  Mulligan,  J.  H.     "Relation  of  Cranial 
Capacity  to  Intelligence,"  /.  Roy.  Anthr.  Inst.,  1923,  53, 

24.  Rogers,  M.   C.     "Adenoids  and  Diseased  Tonsils,  Their 
Effect  on  General  Intelligence,"  Arch.  Psychol.,  1922,  No.  50. 
Pp.  70. 

25.  Sherman,  E.  B.    An  Experimental  Investigation  concerning 
Possible  Correlation  between  Certain  Head  Measurements  and 
University  Grades.    Unpub.  A.B.  Thesis,  Univ.  Wisconsin, 
1923  (also  reported  in  Hull,  12). 

26.  Smillie,  W.  G.,  and  Spencer,  C.  R.  "Mental  Retardation  in 
School    Children    Infested    with    Hookworms,"    /.    Educ. 
Psychol.,  1926,  17,  314-321. 


27.  Sommerville,  R.  C.  "Physical,  Motor,  and  Sensory  Traits," 
Arch.  PsychoL,  1924,  No.  75.  Pp.  108. 

28.  Sterling,  E.  B.,  and  Bell,  E.    "Hearing  of  School  Children 
as  Measured  by  the  Audiometer  and  as  Related  to  School 
Work,"  U.  S.  Public  Health  Reports,  U.  S.  Health  Service, 
1930,45,  1117-1130. 

29.  Terman,  L.  M.    Genetic  Studies  of  Genius.    Stanford  Univ.7 
Calif.:  Stanford  Univ.  Press,  1925.  Vol.  I.  Pp.  648. 


The  relationship  between  mental  and  physical  traits 
has  also  been  considered  from  the  point  of  view  of  con- 
stitutional types.  In  the  effort  to  simplify  the  almost  infinite 
observable  variations  among  individuals,  certain  basic 
human  types  have  been  proposed.  A  specific  individual 
can  then  be  described  as  a  more  or  less  close  approxima- 
tion to  one  of  a  small  number  of  types.  Such  constitu- 
tional types  are  offered  as  a  characterization  of  the  indi- 
vidual as  a  whole,  in  all  his  physical,  intellectual,  and 
emotional  traits,  and  are  not  to  be  envisaged  in  terms  of 
any  isolated  qualities  of  the  organism.  Tl^i^-4s-~alsa»a 

innate  ,or,,  Jxereditary,  ba,sis  to 

the,.jdevelopment,aLj:jpes.  TliusjyJ^         constitutional 

degree  of  conformity  among  the 

various  characteristics,  of  ,/the  individual,  these  character- 
istics beiEtg  ultimately  attributed  to  aa  underlying  innate 

***W**i!i>*W  ,.^, 

Type  theories  have  been  eagerly  seized  upon  by  the 
general  public  as  a  short-cut  to  the  understanding  of 
human  nature.  The  layman  is  impatient  with  the  slow, 
meticulous  methods  of  science.  This  is  particularly  true 
in  psychology,  because  of  the  more  intimate  and  immedi- 
ate bearing  which  this  science  has  upon  man's  everyday 
life.  The  terminology  of  type  theories  has  become  such 

1  The  concept  of  types  has  also  been  employed  in  the  description  of  specific 
functions,  as  in  Galton's  classification  of  individuals  in  regard  to  their  pre- 
dominant field  of  imagery,  i.e.,  visual,  auditory,  olfactory,  etc.  (3).  Such  types, 
however,  do  not  characterize  the  personality  as  a  whole,  and  are  not  to  be  con- 
fused with  the  constitutional  biotypes  under  consideration. 



an  integral  part  of  our  language  that  it  is  almost  impos- 
sible for  us  to  speak  about  people  without  reference  to 
some  hypothetical  categories.  Sooner  or  later  we  inad- 
vertently lapse  into  this  practice.  The  popular  tendency 
to  make  sharp  distinctions,  together  with  the  previously 
discussed  efforts  to  "read"  character  and  mentality  from 
physical  signs,  has  done  much  to  keep  "  types"  alive. 

In  view  of  their  wide  and  persistent  practical  influence, 
type  theories  should  be  closely  examined  and  submitted 
to  experimental  analysis.  Recently  there  has  been  a 
revival  of  interest  in  the  problem  of  constitutional  types 
among  psychologists.  In  the  present  chapter,  we  shall 
survey  briefly  some  of  the  best-known  type  theories, 
inquire  into  their  psychological  implications,  and  exam- 
ine some  representative  data  collected  to  support  or  test 
their  claims. 


The  first  clearly  formulated  attempt  to  classify  indi- 
viduals into  basic  types  was  probably  that  of  the  Greek 

the  fifth  century  B.C.    Hippoc- 

rates proposed  a  two-fold  division  into  habitus  apoplecti- 
cus  and  habitus  phthisicus.    The  former  corresponds  to  a 
thick-set,  heavy  body  build,  susceptible  to  apoplexy  and 
similar  physical  disorders;  the  latter  is  characterized  by  a 
long,  slender  body  and  susceptibility  to  respiratory  dis- 
eases such  as  tuberculosis.    Because  of  the  predominantly 
medical  interest  of  its  exponent,  this  classification  was 
based  primarily  upon  relative  susceptibility  to  different 
kinds  of  physical  ailments.    Sudh^jaiLJ^tpiXia^ 
ever,.  persisted  to,ihe  piesmt^joaanyxurj^^ 
taking,  susceptibility  ,to  ^various  JD^ 

The  second  century  Greek  physician  Galen,  frequently 


called  the  father  of  modern  medicine,  is  reSBQfidlslsJfor 
jthe  jjrdl-J|:j3OT  into  the 

e  .choleric^  ,the^pjbleg»ia"ti%  and  the  melan- 

cholic. These  terms  have  achieved  great  popularity  as 
descriptive  figures  of  speech,  although  one  wonders  how 
often  they  are  still  being  taken  literally.  Tig  theories  of 
bothJHippocrates  ,,.and  ,Galea,4,wer,e,  ,  founded  upon.a(Jbio- 
ch^icaJ^jp^roach^  to  personality.  Thus  Hippocrates 
attributed  the  development  of  his  two  types  to  the  rela- 
tive proportion  of  "fire"  and  "water"  elements  in  the 
individual's  make-up.  Jo^deiLJ^ 


Passing  to  more  modern  times,  we  find  a  wide  variety 
of  type  concepts  in  literature,  art,  philosophy,  medicine, 
anthropology,  or  any  other  field  in  which  man  is  the 
central  figure.  The  English  anthropologist  Walker,1  in 
1852,  wrote  of  "nutritive  beauty,"  "locomotive  beauty," 
and  "mental  beauty."  In  the  following  year,  Carus,1 
a  German  zoologist,  described  three  bodily  types.  These 
were  the  phlegmatic,  in  whom  the  region  of  the  digestive 
organs  is  prominent,  the  athletic,  with  strongly  developed 
bones  and  muscles,  and  the  asthenic,  with  narrow  chest, 
a  long  body,  and  poorly  developed  skeleton  and  muscula- 
ture. In  France,  several  type  theories  have  been  pro- 
posed, chief  among  which  is  that  of  Sigaud  1  and  his 
students.  Sigaud  recognized  four  types,  which  he  desig- 
nated the  digestive,  the  muscular,  the  respiratory,  and 
the  cerebral.  Manouvrier  l  suggested  a  division  into 
makroskele  and  brachyskele,  or  narrow  skeleton  and  broad 
skeleton.  MacAuliffe  1  offered  the  type  plat  (flat)  and  the 
type  rond  (round). 

In  Italy,  Viola  (cf.  19)  formulated  a  theory  which  has 

1  Cf.  Wertheimer  and  Hesketh  (28). 


become  familiar  to  psychologists  through  the  researches 
of  Naccarati  (19,  20,  21)  and  others  in  America.  Viola's 
types  include  the  macrosplanchnic,  the  normosplanchnic, 
and  the  microsplanchnic.  The  macrosplanchnic  individual 
possesses  a  large  trunk  which  is  excessively  developed  in 
comparison  with  the  length  of  the  limbs;  in  such  a  body 
type,  the  horizontal  dimensions  are  relatively  large,  the 
vertical  relatively  small  The  microsplanchnic,  on  the 
other  hand,  has  a  small  trunk  and  long  limbs,  the  vertical 
dimensions  being  relatively  in  excess  of  the  horizontal. 
Between  these  two  extremes  is  the  normosplanchnic,  who 
exhibits  a  proportionate  and  harmonious  development  of 
trunk  and  limbs.  Viola  suggested  a  series  of  body  meas- 
urements to  be  employed  in  classifying  individuals  into 
these  types.  Naccarati  (19)  subsequently  devised  a  mor- 
phologic index  as  a  single  numerical  expression  of  body 
build.  This  index  is  computed  as  follows  : 

f  one  arm  +  length  of  one  leg 

Ayr  T    — 

volume  of  trunk 

The  trunk  volume  is  determined  by  a  series  of  rather 
elaborate  measurements. 

According  to  Viola's  theory,  the  macrosplanchnic  rep- 
resents an  overdevelopment  of  the  nutritional  or  "vegeta- 
tive system"  contained  within  the  trunk.  The  micro- 
splanchnic,  on  the  other  hand,  is  characterized  by  an 
overdevelopment  of  the  "animal  system,"  consisting  of 
the  musculature,  nervous  system,  and  skeleton.  Intel- 
lectual and  emotional  differences  are  attributed  to  the  two 
types,  because  of  the  relative  activity  of  the  vegetative 
and  animal  systems  which  are  regarded  as  independ- 
ent and  even  antagonistic  in  their  action.  The  macro- 
splanchnic  is  regarded  as  representing  a  lower  evolution- 
ary level  than  the  microsplanchnic  because  of  the  greater 


morphologic  resemblance  of  the  former  to  the  newborn 
and  of  the  latter  to  the  adult.  Naccarati  (19),  in  his 
elaboration  of  Viola's  theory,  also  suggests  that  the  micro- 
splanchnic  corresponds  to  a  hyperthyroid  condition  and 
should  therefore  be  expected  to  manifest  the  various 
characteristics  associated  with  overactivity  of  this  gland. 

Pende  (cf.  28)  has  more  recently  proposed  a  classifica- 
tion into  hypervegetative  and  hypovegetative  biotypes, 
which,  as  the  terms  imply,  has  much  in  common  with 
Viola's  theory.  A  definite  endocrine  basis  is  offered  for 
this  theory. 

In  America,  Davenport  (2)   has  classifi^dwj^ 
into  jthe  fleshy,  the  .  mediuii^^ 

'Stockard  (26)  distinguishes  between  the  linear  and  the 
lateral  types,  which  he  ties  up  with  the  activity  of  the 
thyroid.  The  linear  type  is  described  as  active,  energetic, 
and  nervous,  but  emotionally  controlled;  such  individuals 
grow  rapidly  and  reach  puberty  at  a  relatively  early  age. 
The  lateral  type  is  less  active  and  grows  at  a  slower  rate. 
The  former  type  of  individual  is  also  characterized  by  a 
dolichocephalic  skull,  the  latter  by  a  brachycephalic  one. 
Mention  should  also  be  made,  from  the  psychological  side, 
of  the  tiJQaousj^  (8) 


b^anceto  thejntro 

..-ot~tia£u^^  0** 

the  basis  of  observations  made  in  the  course  of  his  condi- 
tioning experiments  on  dogs,  be^jgroposestv^^gre 

I  ntermedi- 
ate,  less  pronounced  types  are  also  described.     Pavlov 


points  to  a  resemblance  between  the  classification  so 
obtained  and  the  classical  division  into  sanguine,  melan- 
cholic, phlegmatic,  and  choleric  temperaments.  He  sug- 
gests that,  "Until  a  rigid  scientific  classification  is  fully 
established  for  all  the  various  types  of  central  nervous 
system  *  .  *  we  may  be  permitted  to  make  use  of  the 
ancient  classification  of  the  so-called  temperaments"  (22, 
p.  286). 

Type  psychology  has  flourished  most  vigorously  in 
Germany.  Numerous  variations  and  ramifications  of 
type  theory  have  been  formulated  by  contemporary  Ger- 
man psychologists.  Weidenreich  (27),  after  a  survey  of 
the  current  type  psychologies,  has  attempted  to  reduce 
all  types  to  two,  namely,  the  leptosome,  with  long,  narrow 
body,  and  the  eurysome,  with  short,  thick-set  body. 
(cf  .  12,  14)  has  proposed  a  classification  qf^con- 

stitutional  types  on  the  basis^&J.^idrtic,,,  imagery.  The 
eidetic  image  is  a  peculiarly  vivid  and  detailed  memory 
image  l  which  is  experienced  by  some  individuals.  Eidetic 
imagery  has  been  found  to  be  most  common  in  late 
childhood  and  to  disappear  gradually  during  adolescence, 
although  it  has  also  been  discovered  among  some  adults. 
The  eidetic  image  may  be  a  photographic  replica  of  the 
original  object,  or  it  may  differ  from  the  latter  in  certain 
characteristic  ways.  Jaensch  and  his  co-workers  maintain 
that  eidetic  images  are  not  a  pathological  phenomenon, 
but  represent  a  normal  stage  in  the  development  of  many 

JaeaadL.i££Og^  eidetic  individuals. 

In  -t4e"~fe«^^  be^ca^ 

The  eidetic  image  "In  '"sucET  cases 

ffi9re.  .jthan  a  .visualized  idea  and  it  is 

1  Eidetic  images  have  usually  been  investigated  in  the  visual  field,  although  it 
has  been  claimed  that  they  are  equally  common  in  other  senses. 


ft^  Individual.   ,  In 

tlxe.jse£ond-type^  the  image,,  naually  ..arises  spontaneously 
and  ,miy.  persevere  in  spite,  oi,  efforts  to  banish  it;  volun- 
tary alterations  in  the  qualities  of  the  image  are  often 
impossible.  Such  images  do  not  come  up  very  frequently 
and  are  often  regarded  as  unpleasant  and  even  uncanny 
by  the  subject.  Jaensch  considers  these  two  eidetic  types  distinct  constitutional  types',  "manifested  in  many 
bodilyLJSSL  mental  traits,_and  characterized  by  basically 
diflFexejg,l-",ps,y:chQphysical  reaction-  systems."  The  eidetic 
characteristics  are  simply  taken  as  convenient  starting 
points  in  the  classification.  Th^-^&^t.-jDf^thjs,  two  ;types 

of  its  alleged  resemblance  to  the  Basedow  ^syndrome,1  and 
the^  second  j  some 

of  its  manifestations  to  the  condition  of-tetany.2 

Jnng!,siatrQv.ert  and  extrovert  tj^pes  are  well-known  (9). 
In  the  extrovert,  the  "  psychic  energy"  is  turned  outward 
to  the  objective  environment.  In  the  introvert,  it  is 
turned  inward  to  a  subjective  world.  The  extrovert  is 
predominantly  oriented  in  all  his  actions,  thoughts,  inter- 
ests, and  feelings,  by  the  objects  and  people  about  him. 
His  beliefs  and  opinions  are  guided  by  the  mores  of  his 
group.  The  introvert,  on  the  other  hand,  is  governed  by 
subjective  factors;  all  his  behavior  has  a  subjective,  inner 
reference.  Jung  regards  these  two  types  as  fundamental 
biological  contrasts.  They  denote  for  him  basic  attitudes 
which  characterize  all  aspects  of  the  individual's  psycho- 
logical make-up. 

1  A  condition  characterized  by  prominence  of  the  eyeballs>  enlargement  of  the 
thyroid  gland,  muscular  tremors,  rapid  heart  action,  and  more  or  less  profound 
mental  disturbance;  believed  to  be  caused  by  overactivity  of  the  thyroid  gland. 

2  A  motor  disorder,  including  muscular  tremor,  muscular  spasms,  and  some- 
times uncoordinated  muscular  contractions  following  upon  an  effort  to  make  a 
voluntary  movement;  attributed  to  insufficient  secretion  of  the  parathyroid 


Jung's  types  have  become  more  widely  known,  how- 
ever, in  terms  of  their  emotional  and  social  manifestations. 
Thus  the  introvert  is  usually  thought  of  as  an  emotionally 
shut-in  individual  who  shuns  social  contacts,  prefers  to 
work  alone,  and  finds  more  pleasure  in  imaginative  work 
than  in  a  life  of  action.  The  extrovert  suggests  the 
"salesman"  type,  who  meets  people  easily,  is  happiest 
in  a  social  situation,  friendly,  and  interested  in  his  fellow- 
beings.  Jung  regards  introversion  and  extroversion  as 
characterizations  of  normal  people.  In  extreme  forms, 
to  be  sure,  they  would  predispose  the  individual  to  mental 
disorders  which  are  opposite  in  their  symptoms.1  The 
fundamental  distinction,  however,  is  not  made  on  the 
basis  of  these  mental  disorders.  The  susceptibility  to  one 
or  another  form  of  insanity  is  considered  simply  one  more 
manifestation  of  the  basic  type. 

Kretschmer's  type  theory  (15)  has  probably  been  the 
most  influential  in  stimulating  research.  Physically, 
^  individuals  into  four  groups,  the 
pmej  wd.  4y.splastic.  The  pyknic 
type  of  body  build  is  short  and  thick-set,  with  relatively 
large  trunk  and  short  legs,  round  chest,  rounded  shoulders, 
and  short  hands  and  feet.  The  athletic  has  a  more  propor- 
tionate development  of  trunk  and  limbs,  well-developed 
bones  and  muscles,  wide  shoulders,  and  large  hands  and 
feet.  The  leptosome  is  generally  characterized  by  small 
body  volume  in  relation  to  height.  He  is  tall  and  slender, 
with  relatively  narrow  chest,  long  legs,  elongated  face, 
and  long,  narrow  hands  and  feet.  In  the  dysplastic  cate- 
gory are  placed  all  individuals  who  present  some  marked 
abnormality  of  physical  development,  disproportion,  glan- 

1This  distinction  has  been  stressed  by  McDougall,  who  states  for  example: 
"  .  .  .  persons  of  the  extrovert  temperament  seem  more  liable,  under  strain, 
to  disorder  of  the  hysteric  or  dissociative  type;  those  of  introvert,  or  shut-in, 
temperament  to  disorder  of  the  neurasthenic  type"  (17,  p.  28). 


dular  imbalance,  or  other  defect.  Kretschmer  suggests  a 
wide  variety  of  physical  measures  to  be  used,  in  conjunc- 
tion with  the  clinical  diagnosis  of  the  experimenter,  in 
differentiating  between  these  bodily  types. 

Th^jDaj^^  .Kmtscbmer'  3  theory  is  th  a  t 

there  exists  a  jrelatiQushipt  between  the  .body  types,  which 
rfe  opposed  "  temperarn^nts," 

^  schizoid.  The  cycloid  individual 
manifests  personality  traits  which  in  extreme  cases  would 
be  classified  under  the  circular,  or  manic-depressive,  form 
of  insanity;  the  schizoid  tends  toward  schizophrenia,  or 
dementia  praecox.  Kretschmer  claims  that  the  cycloid 
is  usually  pykni£^jwhereas  the  TcEzoicT  Is  leptosome  ,pr, 
leslTfrec^^  Although  originally  applied  to 

different  forms  of  mental  disorders,  this  theory  has  sub- 
sequently been  extended  to  normal  individuals  who 
manifest  no  personality  disturbance.  The  terms  cyclo- 
thyme  and  schizothyme  have  been  employed  to  denote 
these  two  normal  biotypes.  The  former  is  described 
as  social,  friendly,  lively,  practical,  and  realistic;  the 
latter  as  quiet  and  reserved,  more  solitary,  timid,  and 
shut-in.  It  will  be  noted  that  these  descriptions  correspond 
closely  to  Jung's  extrovert  and  introvert  types. 

^  jcajnjietect 

a  ggS££§Ldi£]2^ 

types.    From  the  standpoint  of  physique,  the  distinction 

is  one  between  the  long  narrow  body,  with  relatively 
long  limbs,  and  the  short  stocky  build,  with  relatively 
large  trunk  and  short  limbs.  In  respect  to  personality,, 
we  are  offered  at  the  one  extreme  the  expansive,  sociable, 
easy-going,  and  practical  man,  and  at  the  other  the  more 
taciturn,  unsociable,  intellectually  independent,  or  ideal- 
istic type.  Occasionally,  more  than  two  categories  are 
given,  but  the  additional  types  are  usually  found  to  be 


either  intermediate  degrees  or  modifications  of  the  major 

In  some  theories,  the  structural  classification  is  em- 
phasized; in  others  the  behavioral  one  is  foremost.  Many 
of  the  theories  draw  upon  pathological  conditions  either 
for  striking  examples  or  for  their  starting  points,  so  that 
we  frequently  find  susceptibility  to  a  given  class  of  physi- 
cal or  mental  disorders  as  an  outstanding  characteristic 
of  each  type.  In  many  cases,  too,  the  various  physical 
and  personality  types  have  been  linked  to  the  problem 
of  race  and  attempts  have  been  made  to  attribute  racial 
differences  to  the  predominance  of  one  or  another  con- 
stitutional type  within  each  group.  Discussion  of  this 
phase  of  the  problem  will  be  postponed  to  Chapter  XV. 

The  concept  of  types  is  a  broad  one.  We  have  been 
here  concerned  only  with  theories  of  constitutional  bio- 
types  as  described  in  the  opening  paragraph.  For  this 
reason,  no  mention  has  been  made  of  such  theories  as 
that  of  Spranger  (25)^  who  describes  six  fundamental 
types  of  individuality,  namely  the  theoretical  man,  the 
economic  man,  the  aesthetic  man,  the  social  man,  the 
man  of  power,  and  the  religious  man.  These  "types"  are 
regarded  as  meaning-tendencies  or  values  in  terms  of 
which  an  individual's  responses  to  his  environment  are 
to  be  understood.  They  are  ideal  types  or  schemata  of 
understanding,  rather  than  empirically  observable  bio- 


Type  theories  have  been  most  commonly  criticized 
because  of  their  attempt  to  classify  individuals  into 
sharply  divided  categories.  As  was  pointed  out  in  Chap- 
ter II,  such  a  procedure  implies  a  multi-modal  distribution 

1  Cf.  also  the  discussion  by  Kliiver  (13). 


of  traits.  The  introverts,  for  example,  would  be  expected 
to  cluster  at  one  end  of  the  scale,  the  extroverts  at  the 
other  end,  and  the  point  of  demarcation  between  them 
should  be  clearly  apparent.  Actual  measurement,  how- 
ever, reveals  a  uni-modal  distribution  of  all  traits,  which 
closely  resembles  the  bell-shaped  normal  curve. 

Similarly,  it  is  often  difficult  to  classify  a  given  indi- 
vidual definitely  into  one  type  or  the  other.  The  typol- 
ogists,  when  confronted  with  this  difficulty,  have  fre- 
quently proposed  intermediate  or  " mixed"  types  to  bridge 
the  gap  between  the  extremes.  Thus  Jung  suggested  an 
ambivert  type  which  manifests  neither  introvert  nor  ex- 
trovert tendencies  to  a  predominant  degree.  Observation 
seems  to  show,  however,  that  the  ambivert  category  is 
the  largest,  and  the  decided  introverts  and  extroverts  are 
relatively  rare.  The  reader  is  referred,  for  example,  to  the 
distribution  curve  obtained  by  Heidbreder  with  an  intro- 
version questionnaire  administered  to  200  college  stu- 
dents (Ch.  II).  It  will  be  recalled  that  the  majority  of 
scores  were  intermediate  and  that  as  the  extremes  of 
either  introversion  or  extroversion  were  approached,  the 
number  of  cases  became  progressively  smaller.  The  curve, 
too,  showed  no  sharp  breaks,  but  only  a  continuous 
gradation  from  the  mean  to  the  two  extremes.  As  was 
indicated  in  Chapter  II,  the  same  may  be  said  of  all  other 
measurable  traits  of  the  individual,  whether  social,  emo- 
tional, intellectual,  or  physical. 

It  is-^app-a^ei^r^fe^^ 

Such  an  assumption,  however,  is  not  necessarily  inherent 
in  all  systems  of  human  typology.  It  is  more  character- 
istic of  the  popular  versions  and  adaptations  of  type 
theories  than  of  the  original  concepts.  To  be  sure,  type 


psychologists  have  often  attempted  to  categorize  indi- 
viduals, but  this  was  not  an  indispensable  part  of  their 
theories;  their  concepts  have  occasionally  been  sufficiently 
modified  to  admit  of  a  normal  distribution  of  traits. 

Th^,  suggestion  has  been  offered  that  type  theories 
may  refer  simply  to  varieties  or  breeds  of  man,  which 
were  originally  pure  "biotypes."  Through  successive 
generations  of  interbreeding,  mixed  types  have  been  pro- 
duced which  now  outnumber  the  remaining  specimens  of 
pure  types.  It  is  well  known  that  through  the  mechanism 
of  heredity,  interbreeding  will  in  the  long  run  produce  a 
larger  number  of  mixed  than  pure  individuals.  The  same 
applies  to  interbreeding  among  the  proposed  human  bio- 
types.  This  situation  would  then  present  a  normal  dis- 
tribution of  traits,  with  the  largest  number  of  individuals 
in  the  center  of  the  distribution,  corresponding  to  the 
numerically  largest  "mixed"  group.1  Thus  the  form  of 
the  distribution  curve  cannot  in  itself  indicate  the  com- 
position of  the  group.  The  normal  curve  might  be  ob- 
tained with  a  single  intermediate  type  and  minor  devia- 
tions from  it,  or  it  might  result  from  the  mixture  of  several 
pure  biotypes. 

TJie  only  essential  implication  in  this  concept  of  bio- 
types seems  to  be  a  certain  organization  among  the, Car- 
ious characteristics  of  the  individual.  Thus  a  relation- 
ship would  be  expected  between  body  build,  emotional 
reactions,  and  intellectual  traits.  If  there  exist  diverse 
biological  types  of  man,  each  manifesting  its  own  pecu- 
liarities in  physique,  personality,  and  intellect,  we  should 
find  a  certain  degree  of  conformity  among  these  char- 
acteristics of  the  individual.  Wfeea.  &Q,.coacdXQd^,^the 
problem  of  types  is  ultimately  reducible  to  a  considera- 

1 A  particularly  lucid  statement  of  this  point  of  view  can  be  found  In  Klineberg 
etal.  (ID).' 


tion  of  the  relationship  between  .structural  and  behavioral 
qualities.  It  is  not,  however,  concerned  with  isolated 
traits,  but  with  the  composite  picture  of  the  individual 
as  a  whole. 

Insofar  as  the  existence  of  human  biotypes 
determined  from  the  form  of  the  distribution  curve,  the 
question  may  be  raised  regarding  available  techniques  of 
typological,  investigation.  One  method  involves  the  classi- 
fication of  individuals  into  distinct  groups  on  the  basis  of 
personality  tendencies,  and  the  subsequent  comparison 
of  these  groups  in  regard  to  physique.  This  technique  has 
been  employed  largely  with  abnormal  cases  in  the  effort 
to  check  the  assertions  that  a  given  physique  predisposes 
the  individual  to  a  certain  kind  of  mental  disorder.  Thus, 
for  example,  the  relative  number  of  pyknics  and  lepto- 
somes  among  individuals  manifesting  different  forms  of 
insanity  has  been  compared  and  evaluated  in  terms  of  the 
expected  association. 

A  second  method  is  based  on  the  correlational  analysis 
of  measurements  collected  on  large  normal  groups.  Vari- 
ous physical  indices  of  body  build  have  been  worked  out 
for  this  purpose.  Such  indices  are  then  correlated  with 
test  scores  or  ratings  on  crucial  personality  traits.  A 
high  correlation  would  be  evidence  for  the  conformity 
implied  by  type  theories. 

Finally,  efforts  have  been  made  in  a  few  studies  to 
identify  and  select  "pure  types"  and  then  investigate 
thoroughly  the  characteristics  of  these  individuals.  The 
types  are  chosen  on  the  basis  of  physical  characteristics, 
so  as  to  represent  "good  specimens"  in  this  one  respect. 
The  physically  contrasting  groups 'are  then  compared  in 
emotional  and  intellectual  reactions.  This  method  rests 
upon  two  questions.  First,  can  individuals  be  found  who 
correspond  to  the  alleged  biotypes  in  all  their  physical, 


intellectual,  and  emotional  characteristics?  Secondly, 
when  persons  are  put  into  classes  on  the  basis  of  clear- 
cut  and  extreme  physical  diversities,  will  they  also  exhibit 
corresponding  differences  in  other  traits?  It  will  be  noted 
that  the  second  question  is  similar  to  that  which  underlies 
the  study  of  types  among  the  psychotic  and  neurotic. 
In  the  present  method,  however,  the  classification  is 
employed  on  normal  rather  than  on  pathological  cases, 
and  physical  rather  than  emotional  traits  are  taken  as 
the  starting  point.  Some  of  the  major  investigations  by 
each  method  will  be  surveyed  in  the  following  sections. 

Kretschmer  originally  formulated  his  theory  of  consti- 
tutionai  types  from  observations  on  psychotic  patients, 
In  comparing  the  body  build  of  schizophrenics  and  manic- 
depressives,  he  has  consistently  found  a  greater  propor- 
tion of  leptosomes  among  the  former  and  pyknics  among 
the  latter.  Recently  Kretschmer  (16)  has  compiled  data 
from  several  investigators  on  over  4000  abnormal  cases, 
with  the  following  results  : 

Body  Type  Schizophrenic  Manic-Depressive 

Pyknic  and  mixed  pyknic  12.8%  66.7% 

Leptosome  and  athletic  66.0  23  .6 

Dysplastic  11.3  °-4 

Unclassifiable  9-9  9-3 


.and  athletic  categories, 

SS^  E££centa?e  °^  ,raanic-cl,epressivea  fall 


(28)  measured  65  male  pa- 

tients chosen  at  random  from  two  American  institutions 
for  the  insane.    Of  these,  n  had  been  clearly  diagnosed 


as  manic-depressive  and  23  as  schizophrenics.  The  major 
part  of  the  investigation  was  therefore  confined  to  these 
cases.  The  subjects  were  first  classified  into  Kretschmer's 
body  types  on  the  basis  of  general  observation.  A  series 
of  53  anthropometric  measurements  were  then  taken  and 
various  bodily  indices  computed.  One  of  these  indices  was 
ultimately  selected  as  the  most  satisfactory  l  and  adopted 
as  the  chief  basis  of  classification.  A  close  correspondence 

"~^--»-»»B»«U*,HK          J.UBHJ./F,."     u*     JCT,.  ,    ,  (<  , 

was  found  between  the  two  procedures,.  Those  individuals 
classified  ^  as  pyknic  by  the  experimenter's  diagnosis  invari- 
ably .haAindices  under  255;  those  classified  as  leptosomes 
had  indices  over  270.  There  was  no  overlapping  in  the 
indices  of  these  two  groups.  JBjLeWi^nig^gdjpf  classifi- 
catonj_haK,ffifir^  the  Dumber  pl,d,ecided.,pyknics_or  lepto- 
some^was  small^j^Q^t  ,,  individuals  f  ailing  ,iatQ  ,  the  rinter- 
mediate^athletic^Qr^mized,  groups,  a.s  would  be  expected. 

The  percentages  of  persons  of  each  body  type  found 
in  the  schizophrenic  and  manic-depressive  groups  are 
given  below: 

Body  Type  Schizophrenic  Manic-Depressive 

(N  =  23)  (N  «  n) 

Pyknic  4.3%  45-5% 

Pyknoid  13.0  36.4 

Athletic  26.1  9.0 

Leptosome-athletic-mixed  34.8  o. 

Leptosome  17.4  o. 

Unclear  4.3  9.0 



scatter  over  a  wider  variety  of  body  type,  but  the  greatest 

**V"****w*«M|«M*U^«,  <>«*'»*   'a*J**'l^l"*'V'''^''1   '      r'  '          "         1P'"    ''*  "-H'l   i  P/    'S»*W«<»*4W**1*f*  ''*   V***'  *'»««««''  S-El*****1"*1 

,aad  athktip, 

,     _  IPO  X  leg  length  X  io3 


transverse  chest  diameter  X  sagittal  chest  diameter  X  trunk  height 
(cf.  28,  p.  415). 


The  chief  difficulty  in  interpreting  the  results  of  these 
and  similar  investigations  on  psychotic  cases  is  inadequate 
control  of  the  age  factor.  Schizophrenia  is  more  common 
among  younger  subjects,  whereas  older  people  are  more 
susceptible  to  manic-depressive  psychoses.  It  is  also  a 
well-established  fact,  which  Kretschmer  himself  recog- 
nizes, that  older  subjects  tend  more  toward  the  pyknic 
body  build,  younger  subjects  toward  the  leptosome.  To 
be  sure,  pyknics  may  be  found  among  young  people, 
and  leptosomes  among  older  groups;  and  many  individuals 
retain  the  same  type  of  body  build  throughout  life.  But 
the  general  trend  is  sufficiently  marked  to  produce"  aft 
entirely  spurious  relationship  between  body  build  "and 
psychotic  tejidencies.  For  this  reason,  it  is  essential  that 
age  differences  be  ruled  out  in  any  comparison  of  the 
body  type  of  different  psychotic  groups. 

In  a  recent  investigation  by  Garvey  (5),  130  manic- 
depressives  and  130  dementia  prsecox,  or  schizophrenic, 
patients  were  selected  so  that  the  two  groups  would  be 
closely  matched  in  age.  Only  clear  cases,  classified  with 
complete  agreement  by  the  hospital  staff  (not  including 
the  experimenter),  were  employed.  When  the  patients 
were  divided  into  heavy  and  slender  types . on  the  basis 
of  n  general  observation,  some  evidence  was  found  for 
Kretschmer's  claims.  The  association,  however,  is  re- 
ported as  too  slight  to  permit  body  type  to  be  regarded 
as,  diagnostic  of  psychosis.  Extensive  physical  measure- 
ments were  taken  and  several  ratios  between  horizontal 
and  vertical  bodily  dimensions  were  computed.  All  showed 
an  almost  complete  overlapping  of  the  two  psychotic  groups. 
Not  only  were  the  averages  closely  similar,  but  the  range 
and  the  general  form  of  the  distribution  were  practically 
identical  in  the  two  groups. 

Naccarati   (20),  in  the  effort  to  check  upon  Viola's 


hypothesis,  measured  100  male  Italian  psychoneurotics 
between  the  ages  of  25  and  40.  The  number  of  normo- 
splanchnics  is  reported  as  being  smaller  in  this  group  than 
in  normal  groups.  The  neurasthenics  had  a  larger  propor- 
tion of  microsplanchnics  (long,  slender  type),  while  macro- 
splanchnics  predominated  among  the  "emotional  psycho- 
neurotics."  Under  the  latter  category  Naccarati  includes 
cases  of  hysteria,  anxiety  neuroses,  and  traumatic  neu- 
roses. Averages  of  some  of  the  most  significant  physical 
measurements  as  well  as  the  average  age  of  the  two 
groups  are  given  below. 

Q  Morphologic      Total  Volume     Length  of          * 

Index  of  Trunk         Extremities 

50  Neurasthenics  456.64  3°-43  I33-35        32.16 

50  Emotional  psycho- 
neurotics  362.06  37-36  128.80        33-94 

It  will  be  noted  that  the  neurasthenic  group  has  a  lower 
average  age  than  the  emotional  psychoneurotics.  This 
might  account  in  part  for  the  greater  tendency  to  micro- 
splanchny  among  the  former.  No  account  is  given  of  the 
method  of  obtaining  or  diagnosing  the  subjects,  a  fact 
which  makes  interpretation  of  the  findings  difficult. 

Although  not  dealing  with  abnormal  patients,  a  subse- 
quent investigation  by  Naccarati  and  Garrett  (21)  may 
be  mentioned  in  this  connection.  The  subjects  were 
54  male  college  students  between  the  ages  of  18  and  25. 
The  general  approach  was  similar  to  that  of  the  other 
investigations  reported  in  this  section.  The  subjects 
were  classified  into  three  groups  on  the  basis  of  number 
of  neurotic  symptoms  indicated  on  the  Woodworth  Per- 
sonal Data  Sheet.1  The  investigators  also  obtained  ratings 

1  The  students  were  also  given  the  Pressy  X-O  and  the  Downey  Will-Temper- 
ament tests,  but  the  results  were  difficult  to  interpret,  probably  because  of  the 
inadequacy  of  these  tests  when  applied  to  college  groups. 


by  the  class  instructor  and  self-ratings  of  each  student 
on  physique  (including  bearing,  apparent  health,  vigor, 
strength,  and  endurance),  intelligence,  emotional  stabil- 
ity, and  aggressiveness.  Morphologic  indices  were  com- 
puted for  each  subject.  The  average  Woodworth  P.D. 
score  for  each  group  is  shown  below,  together  with  the 
averages  and  S.D.'s  of  the  morphologic  indices.  It  will 
be  recalled  that  a  high  morphologic  index  indicates  micro- 
splanchny,  a  low  morphologic  index  macrosplanchny. 

Group  I         Group  II       Group  III 
(N  =  18)       (N  =  16)       (N  =  20) 

Woodworth  P.D.  Score:  Average  6.9  14.2  24.8 

Morphologic  Index:  Average  49&-S  495 -°  460.8 

Morphologic  Index:  S.D.  61.3  57.5  56.0 

It  will  be  noted  that  group  III,  composed  of  the  most 
neurotic  individuals,  has  a  much  lower  average  morpho- 
logic index  than  the  other  two  groups.  The  self-ratings 
and  instructors'  ratings  agreed  closely  with  each  other 
and  with  the  test  results.  The  average  differences  among 
the  three  groups  were  slight  but  in  the  expected  direction. 
Thus,  in  emotional  stability,  group  I  was  rated  as  the  most 
stable  and  group  III  as  the  least  stable.  Similarly,  group  I 
was  rated  as  most  intelligent,  group  III  least  intelligent. 
These  data  seem  to  support  Viola's  contentions  in  a 
general  way.  The  macrosplanchnics  appear  less  intelligent 
and  have  a  more  pronounced  tendency  to  neuroticism 
than  the  microsplanchnics.  Several  reservations  should, 
however,  be  borne  in  mind.  The  groups  are  small;  age 
may  have  differed  appreciably  among  the  three  categories; 
and  finally,  much  overlapping  in  morphologic  index  k 
indicated  by  the  large  S.D.'s.  Variations  in  body  build 
within  each  group  were  larger  than  the  differences  between 
the  averages. 


Aj^y_cpmp,i:ehensive  investigation  on  the  relationship 
^L^SfJ1  body  $7P5  a?*d  Ps7c^otic  disposition  was  recently 
QSS^^  jBur chard  (i).  A  total  of  407  male  White 
patients  from  several  institutions  for  the  insane  were 
selected  for  the  survey.  Of  these,  125  were  clearly  diag- 
nosed as  schizophrenes  by  the  hospital  staff,  and  125  as 
manic-depressives.  The  remaining  157  patients  mani- 
fested a  variety  of  psychotic  and  neurotic  conditions  and 
were  employed  as  a  control  group.  The  subjects  in  all 
three  groups  were  classified  into  pyknics,  athletics,  and 
leptosomes  by  "general  impression."  Comparisons  were 
also  subsequently  made  in  respect  to  several  anthro- 
pometric  measures  and  indices.  Only  seven  dysplastics 
were  found  in  the  entire  sampling,  and  these  were  elim- 
inated from  further  consideration.  All  other  subjects 
were  retained,  any  intermediate  or  mixed  types  being 
assigned  to  the  morphological  type  which  they  resembled 
most  closely.  Below  are  given  the  percentages  of  pyknics, 
athletics,  and  leptosomes  found  in  the  manic-depressive, 
schizophrenic,  and  control  groups,  respectively,  when 
the  "impressionistic"  method  of  classification  was  em- 

Morphological  Type     Manic_Depressives     Schizophrenes          Control 

Pyknic  63.2  36.3  55.6 

Athletic  8.8  17.7  "  11.3 

Leptosome  28.0  46.0  33.1 

The  general  to^j^J^  *n  agree- 

mu^^pTy.  ,Not  only  are  the  greatest 
percentage  of  manic-depressives  pyknics,  and  the  greatest 
percentage  of  schizophrenes  leptosomes,  but  the  control 
group  occupies  a  position  intermediate  between  these  two 
groups  in  all  percentages.  When  the  schizophrenes  and 



manic-depressives  are  compared  in  terms  of  anthropo- 
metric  measures,  a  certain  amount  of  differentiation  is 
also  revealed.  Perfectly  reliable  differences  between  the 
averages  of  the  two  groups  were  found  in  three  out  of 
nine  physical  measures  and  in  two  out  of  three  bodily 
indices.  The  overlapping  of  the  groups  in  all  of  these 
treasures  3^as  nevertheless  very  large.  This  is  illustrated 
in  Figure  36,  which  shows  the  frequency  distributions  on 

Manic -Depressive 

Dementia  -Praecox 

130    150    170    190    210    230    250    270    290    310    330    350     370    390 

Index   Value 



(After  Burchard,  I,  p.  47,) 

Wertheimer-Hesketh  index  of  body  build  (cf.  above). 
This  index  yielded  the  largest  differences  between  the 
two  groups.  It  is  apparent  that,  in  spite  of  the  statisti- 
cally significant  differences  in  averages,  schizophrenes  can 
be  found  who  are  much  more  pyknic  than  certain  manic- 
depressives,  and  vice  versa. 

Even  the  differences  in  averages  between  the  two  groups 
may  be  owing  to  other  factors  which  have  not  been  con- 
trolled. Burchard  recognized  this  difficulty  and  under- 
took a  detailed  analysis  of  his  •  manic-depressive  and 
schizophrene  groups.  In  regard  to  racial  and  national 
background,  occupation,  and  educational  status,  no  ap- 
preciable or  consistent  differences  could  be  discovered. 
In  age,  however,  the  differences  were  very  large,  the  aver- 
age ages  of  schizophrenic,  control,  and  manic-depressive 


groups  being  30.97,  42.90,  and  49.65  years,  respectively. 
Further  analysis  revealed  a  definite  relationship  between 
age  and  body  build.  This  factor  seems  to  have  accounted 
largely,  although  not  entirely,  for  the  group  differences 

Since  the  age  factor  plays  such  an  important  part  in 
all  studies  on  constitutional  type,  we  may  examine  more 
closely  Burchard's  data  on  this  problem.  Below  will  be 
found  the  average  Wertheimer-Hesketh  indices  of  sub- 
jects falling  in  successive  decades  within  the  entire  sam- 
pling as  well  as  within  each  psychotic  group. 

Average  Wertheimer-Hesketh  Indices 
**e        Entire  Group      Schizophrenes      Manic-Depressives       Control 

15-19  306,11  297.25  262.66  321.00 

20-29  275.10  279.77  252.00  273.48 

30-39  260.82  272.00  256.33  253.86 

40-49  249.34  252.50  246.52  249.41 

50-59  253.68  277.50  247.29  257.16 

60-69  236.50  243.33  241.67  228.75 

These^jtverag^g^  indicate,  -  a. .  definite  -teBdency  towards »  a 
more  pyknic^ body  build  with  „  advancing  .age.  JJlus, jg, 
sfeH^  within  each  dis£&S£Jg2^^  the 

Further  corroboration  of  this  finding  is 
furnished  by  the  correlation  of  —.256  obtained  between 
age  and  index  value  in  the  entire  sampling.  Much  of  the 
difference  observed  between  the  two  psychotic  groups 
can  therefore  be  attributed  to  age.  It  should  be  noted, 
however,  that  within  each  decade  the  schizophrenes  have 
a  higher  average  index  than  the  manic-depressives.  To  be 
sure,  the  differences  are  considerably  reduced  by  ruling 
out  age,  and  the  control  group  no  longer  retains  its  inter- 
mediate position,  but  a  certain  difference  in  the  expected 
direction  remains.  Burchard  himself  is  very  cautious  in 
interpreting  these  findings,  concluding  ".  .  .  that  age 


is  a  very  important  conditioning  factor  upon  the  mor- 
phological habitus  and  that— although  the  type  of  psy- 
chosis apparently  play(s)  an  important  role — the  relative 
weights  of  the  two  factors  remain  in  considerable  doubt" 
(i,  p.  65). 


It  has  frequently  been  objected  that  one  cannot  gen- 
eralize from  a  slight  correspondence  between  body  build 
and  certain  forms  of  insanity  to  a  relationship  between 
personality  traits  and  bodily  characteristics  of  normal 
individuals.  Thes>conj,p,gjlsan Htgi , average  values,  further- 
more, or  of  the  percentage  frequency  of  bodily  types 
among;  different  .groups  may  exaggerate  a  very  slight 
degree  of  association.  Such  comparisons  tell  us  little 
abc^jidiyidual  cases.  For  these  reasons,  a  number  of 
investigators  have  resorted  to  the  correlation  coefficient 
to  pfitain  an  exact  quantitative  measure  of  the  amount  of 
relationship  within  a  group. 

The  correlation  coefficient  is  affected  not  only  by  the 
presence  or  absence  of  clear-cut  types  within  a  group,  but 
also  by  the  degree  to  which  a  given  typal  characteristic 
is  exhibited.  This  method  seems  to  rest  upon  a  slightly 
different  principle  than  that  underlying  group  compari- 
sons. Thus  if  morphological  index  were  found  to  corre- 
late highly  with  intelligence,  it  would  mean  not  only  that 
the  clearly  microsplanchnic  are  more  intelligent  than  the 
clearly  macrosplanchnic,  but  also  that  within  the  inter- 
vening range,  the  more  microsplanchnic  the  individual, 
the  more  intelligent  he  will  be.  A  lack  of  relationship 
between  intelligence  and  body  build  within  the  inter- 
mediate mixed  groups  will  considerably  lower  the  corre- 
lation which  would  be  obtained  if  only  "pure  types" 
were  included. 


Naccarati  (19)  found  a  correlation  of  ,356  between 
morphologic  index  and  Thorndike  Intelligence  Examina- 
tion for  High  School  Graduates  within  a  group  of  75  col- 
lege men.  In  the  same  study,  height-weight  ratios  l  were 
computed  for  221  college  men  ranging  in  age  from  17 
to  22,  all  of  whom  had  taken  the  Thorndike  test.  The 
correlation  between  this  ratio  and  intelligence  proved  to 
be  .230. 

Ihe^^light  positive  correlation  between  height-weight 
ratio  and  in;^  support  the  claim 

tha4Jiie.mierQ&pl^^  individual,  tends 

to  be  more  intelligent.  Tti£L-age  factor,  however,  must 
again^be^  considered.  Upon  further  statistical  analysis  of 

the  data,  ,  it  wa&.jiijB£az^  of  -230 

r&gulled,  largely  frojr^a  negative  correlation  between  weight 
and  intelligence  ..test,  score  within  this  group.2  The  more 
heavily  built,  stocky  iadiYiduals  at  the  age  levels  under 
consideration  tend  to  be  t;he,j9/^r  members  of  the  group. 
Sinyjji^  .any  one  academic 

lev/l^art  usn&^  It  therefore  .seems  very 

likely  that  even  the  low  degree  of  correspondence  found 
between  height-weight  ratio  and  intelligence  is  attributable 
to  an  uncontrolled  age  factor  and  cannot  be  accepted  as 
proof  of  a  relationship  between  body  type  and  mentality. 
Subsequent  investigations  by  a  similar  method  have 
likewise  cast  doubt  upon  Naccarati's  conclusions.  Heid- 
breder  (6)  checked  Naccarati's  hypothesis  on  a  group 
of  1000  White,  native-born  college  freshmen,  consisting 
of  500  men  and  500  women.  Scores  on  the  Minnesota 


1  The  height-weight  ratio  has  frequently  been  substituted  for  the  more  elabo- 
rate morphologic  index,  for  the  sake  of  expediency,  since  the  two  indices  are 
closely  related.  Naccarati  (19),  for  example,  found  a  correlation  of  .70  between 
the  two  in  a  group  of  75  students,  and  a  correlation  of  .75  in  another  group  of  50. 

2  Subsequently  computed  by  Hull  (7,  pp.  142-143),  from  Naccarati's  published 


College  Ability  Tests  were  correlated  with  the  height- 
weight  ratios  obtained  by  careful  measurements  on  each 
subject.  This  correlation  proved  to  be  only  .03  in  the 
group  of  men  and  .04  in  the  women's  group.  Similarly, 
the  correlation  between  height-weight  ratio  and  scores 
on  each  of  the  five  subtests  of  the  intelligence  exami- 
nation closely  approximated  zero,  ranging  from  —.07 

Irkljie  effort  to  discover  whether  the  use  of  the  more 
elaborate  morphologic  index  in  place  of  the  height-weight 
ratio"  "might  yield  more  positive  evidence  for  Naccarati's 
view,  Sheldon  (23)  conducted  an  elaborate  investigation 
on  434  freshman  men.  All  subjects  were  White;  age  var- 
ied between  17  and  22.  Twelve  measurements  were 
carefully  made  on  each  individual  and  from  them  were 
computed  the  morphologic  indices,  in  the  manner  de- 
scribed by  Naccarati.  The-  correlation  between  these 
indices  and  scores  on  a  common  group  intelligence  test 
^2L59^?^  'freshmen  was  .14.  Correlations  of  the  morpho- 
logic indeFwitE'eacK'oF'tlie  nine  subtests  in  the  examina- 
tion ranged  from  —  .02  to  +.12.  These  ^findings  cor- 
roborate closely  those  obtained  by  Heidbreder  with  the 
heigE-weight  ratio. 

^  morphologic  .types,  Sheldon 

(24)  correlated  ^morphologic  index  and^xatings  oa  .five 
peisoaalityiJrcLlts  withia  a,  group-oi  Z5|  Jrsshmaa  .  men. 
Each  student  was  rated  by  five  upperclassmen  who  be- 
longed to  the  same  fraternity  as  the  subject.  The  judges 
had  thus  had  considerable  opportunity  to  observe  the 
student's  everyday  behavior  in  many  situations  and  were 
fairly  well  qualified  to  rate  him.  The  consensus  of  all 
five  judges  was  taken  as  the  final  rating  for  each  individual. 
Below  will  be  found  the  correlations  between  morphologic 
index  and  ratings  on  each  trait, 


Correlation  with 
Trait  Rated  Morphologic  Index 

Emotional  excitability  .00 

Aggressiveness  —  .08 

Leadership  —.14 

Sociability  —  .22 

Perseverance  .01 

On  the  whole,  these  correlations  are  too  low  to  indicate 
anji^greciable  degree  of  relationship  between  bodily  type 
aniijgersonalitj  traits.  The  correlations  of  morphologic  in- 
dex with  leadership  and  sociability  are,  however,  sugges- 
tive. These  two  correlations  indicate  a  tendency  for  the 
more  heavily  built  individual  to  be  more  sociable  andjoaore 
6Fa  leader.  This  may  again  be  owing  to  a  disturbing  age 
factor,  inasmuch  as  the  older  individuals  within  such  a 
group  might  well  be  expected  to  manifest  these  character- 

both  intel- 

lectucd  jLndjjersonality  .traits  was  conducted  by  Garjrett 
The  subjects  were  again  male  college 

freshmen.  Morphologic  indices  were  computed  with  meas- 
urements taken  from  three  standard  photographs  of  each 
subject.  These  photographs,  taken  in  connection  with 
gymnasium  routine,  showed  three  different  views  of  the 
individual  in  the  nude.  The  morphologic  indices  computed 
from  the  photographs  correlated  .81  with  height-weight 
ratios  obtained  from  direct  measurements  on  219  stu- 
dents. On  this  basis,  the  authors  felt  justified  in  their 
use  of  the  photographs  for  the  sake  of  expediency.  JThe 


the  George  Washington  Social  intrfKgSficfe  Test, 

with  the  following  results  : 

"»»,.,,„    .  ......  -j^jadwiuiuju       '•<*«•"  '*""'  "*"       "     ' 



,    , 


Thorndike    Intelligence 


Woodworth  P.D.  Sheet 
Social  Intelligence  Test 


r  n 
of  Cases 



-       .    . 




Height-Weight  Ratio 
(fmn  ^^  neasure, 



f  n 
of  Cases 

_       .    . 



f.  these  ^correlations  is  sufficiently  large  to  indicate 
a  significant  degree  of  relationship,. 


The  study  of  constitutional,,  types  through  the  correla- 
tion technique  has  yielded  consistently  negative  results. 
As_  has  already  been  pointed  .put,  „  .however,  this  may  be 
owing  to  the  presence  of  a  predominantly  Urge,  group  of" 
mixed  types  Jn  jrhoja  ,iw,,CQ]a^sten^ 

and  npLental  or  emationaL  traits  isi.lo^.iQvjid. 

It  has  also  been  argued 

that  even  when  indices  are  employed  in  lieu  of  isolated 
dimensions,  the  investigator  is  not  getting  a  picture  of 
the  individual's  physique  in  its  totality.  And  the  latter 
is  essential  in  any  concept  of  constitutional  types. 

Most  of  the  numerous  German  investigations  on  types 
have  proceeded  by  selecting  good  specimens  of  each  type 
on  the  basis  of  physical  measurements  or  observations 
and  then  administering  a  variety  of  psychological  tests 
to  the  groups  so  obtained.  By  this  method,  for  example, 
the  conclusions  have  been  reached  that  pyknics  are 
more  distractable  than  leptosomes,  that  they  have  a 
greater  perception  span,  a  better  incidental  memory,  re- 
spond "synthetically"  rather  than  "analytically"  to  a 
difficult  perception,  are  more  sensitive  to  colors  than  to 


forms,  are  superior  in  motor  tasks  except  when  these 
require  delicacy  of  movement,  and  give  more  extroverted 
responses.  These  are  among  the  major  differences  reported 
by  German  investigators.1  Relatively  little  stress  is  placed 
by  the  latter  upon  differences  in  general  intelligence  be- 
tween the  types. 

Many  of  these  studies  are  open  to  serious  criticism 
and  it  is  therefore  difficult  to  evaluate  their  findings. 
The  groups  employed  are  frequently  small 

arej^eported  with  no  indication  of  variability^within  each 
group  or  of  amount  of  o^^a^mgbttween  grouPs«  4Q.Ean" 
titative  data  ^etfreguently  lacki,iiig  and  only  descriptive 
observations  reported.  TQ^tfeSiS^  or 

poorh^standardized.     The 

chieflyjon  the  basis  .of  ,  physical^  in 

other  essentiaL  respects.  Thus  the  relative  proportion  of 
men'^ndjv'omen  may  not  have  been  constant  in  all  the 
groups.  The  pyknics,  furthermore,  may  have  ,be,en,  older 
than  the  leptosomes,  and  this  age  difference  could  account 
for  J^lgbserved  psychological  differences.  Little  or  no 
attempt  has  been  made  to  control  this  factor,  in  some 
studies  the  subjects  ranging  from  adolescents  to  sexage- 
narians. SociaLaJLi*^^ 

affected  the  results.  There  is  some  evidence,  for  example, 
thaFTeptosoTnes  are  found  more  commonly  among  the 
higher  social  and  educational  levels.  Since  there  are  also 
intellectual  and  possibly  emotional  differences  among  these 
classes,  the  factor  of  social  level  should  be  ruled  out. 

Mohr  and  Gundlach  (18)  conducted  an  intensive  quanti- 
tative investigation  on  a  group  of  male  convicts  in  the 
Illinois  State  Prison.  A  total  of  600  men  was  measured, 
out  of  which  89  were  selected  as  good  representatives  of 
leptosome,  athletic,  and  pyknic  types.  In  arriving  at  this 

1  For  a  survey  of  many  of  these  investigations,  see  Klineberg  et  al.  (10). 


classification,  the  investigators  employed  all  the  anthro- 
pometric  measures  suggested  by  Kretschmer,  as  well  as  a 
general  observational  diagnosis  of  body  type.  Each  sub- 
ject was  given  the  Army  Alpha  as  well  as  about  a  dozen 
simple  psychological  tests  suggested  by  the  German  work- 
ers as  diagnostic  of  constitutional  type.  Such  tests  were 
included  as  speed  of  tapping  and  of  writing,  visual  reac- 
tion t  time,  cancellation,  substitution,  color  fusion,  Ror- 
schatch  ink  blots,  etc. 

A  striking  difference  in  average  Army  Alpha  score  was 
found  among  the  three  groups.  This  is  shown  below,  together 
with  the  number  of  cases  in  each  group  and  the  average  ages. 

Body  Type  Number  of  Cases        Average  Age  Alsbh    S  T    ^ 

Leptosome  19  28.55  96.5 

Athletic  26  28.65  79-2 

Pyknic  44  34.75  57.9 

The  correlation  between  Army  Alpha  score  and  an  index 
of  body  build  was  found  to  be  —.34,  which  further  cor- 
roborates the  above  results.  Although  not  very  high, 
this  correlation  indicates  a  significant  tendency  for  the 
tall,  slender  individuals  to  obtain  higher  scores.  Similarly, 
in  many  of  the  other  tests  the  differences  among  the 
groups  were  large  enough  to  be  statistically  significant. 
It  will  be  noted,  however,  that  there  is  a  marked  differ- 
ence in  age  among  the  three  groups,  the  pyknics  being 
on  the  average  a  little  over  six  years  older  than  the 
leptosomes  or  athletics.  In  view  of  the  tendency  for  Army 
Alpha  scores  to  decrease  with  age  (cf.  Ch.  VII),  there- 
fore, the  pyknic  group  would  be  expected  to  obtain  lower 
scores.  The  cultural  and  racial  composition  of  the  three 
groups  is  not  stated,  and  these  factors  may  also  account 
for  some  of  the  observed  differences  in  test  performance. 
More  recently,  Klineberg,  Asch,  and  Block  (10)  have 


attempted  to  compare  Kretschmer's  body  types  under 
more  rigidly  controlled  conditions.  Thejstudy  was  limited 
exclusively  to  college  students,  so  that  variations  in  age 
and  in  social  and  educational  level  were  largely  reduced. 
The  first  group  was  composed  of  153  male  college  stu- 
dents, with  an  average  age  of  19  years-9  months.  All 
were  attending  the  same  institution  and  were  very  homo- 
geneous in  respect  to  racial  and  cultural  background. 
The  subjects  were  classified  into  leptosome  and  pyknic 
categories  on  the  basis  of  experimenter's  diagnosis  as 
well  as  five  indices  computed  from  anthropometric  meas- 
urements. Each  subject  was  given  the  Otis  Self-Adminis- 
tering test  of  intelligence,  an  emotional  adjustment  test, 
and  a  series  of  six  tests  designed  to  measure  alleged  char- 
acteristics of  the  two  opposed  constitutional  types. 

It  was  possible  to  obtain  two  groups  which  were  clearly 
differentiated  in  respect  to  physique.  This  is  illustrated 
by  the  frequency  distribution  curves  of  the  leptosome  and 
pyknic  groups  in  Pignet  Index,1  reproduced  In  Figure  37. 


Pyknic  <N=  56) 

—  Uptosome  (N=59) 





l     I     I     I     I     I     I     t     II 

-15     -10      -50         5        10       15       20       25       30       35       40 

Pignet   Index 

PIGNET  INDEX.    (After  Klineberg,  Asch,  and  Block,  10,  p.  180.) 

1  Pignet  Index  =  Height  —  (weight  +  chest  circumference). 

NOTE:  This  formula  is  printed  incorrectly  in  the  study  under  consideration 
( n,  p.  164).  We  assume  this  was  a  misprint,  and  that  the  correct  formula  was 
employed  in  the  computations. 


This  index  was  found  to  differentiate  more  clearly  between 
the  groups  than  any  of  the  other  physical  measures  em- 
ployed. It  will  be  noted  that  overlapping  is  virtually 
absent.  In  sharp  contrast  to  this  distribution  is  that 
given  in  Figure  38,  which  shows  the  scores  of  leptosomes 



«. —  Leptosome  (N=57) 



z  8 

24  32  40          48  56  64  72  80 

Number  of  Letters  Cancelled 

CANCELLATION.   (After  Klineberg,  Asch,  and  Block,  10,  p.  180.) 

and  pyknics  in  one  of  the  psychological  tests,  viz.,  can- 
cellation of  letters.  In  this  case,  the  two  groups  overlap 
almost  completely.  Similar  results  were  obtained  with 
all  of  the  other  tests.  ]jEUULJS&$^^ 

tion of  measures  on  no  cases  confirmed  these  findings. 
The  correlations  between  physical  indices  and  test  scores 
were  all  close  to  zero.  Intercorrelations  of  the  various 
psychological  tests  were  also  negligible.  If  the  under- 
lying conformity  implied  by  type  theories  were  present, 
a  fairly  close  correspondence  should  have  been  found 


among  the  various  diagnostic  tests.  Viewed  from  any 
angle,  the  results  are  completely  negative. 

A  parallel  study  was  conducted  on  79  women  students 
in  one  college,  all  of  whom  were  between  the  ages  of  16 
and  20.  Various  physical  measures,  as  well  as  scores  on 
the  Scholastic  Aptitude  Test  were  already  available  on 
these  subjects.  They  were  given,  in  addition,  an  emotional 
adjustment  test  and  three  simple  tests  suggested  by  the 
German  type  studies.  It  proved  impossible  to  classify 
the  subjects  into  bodily  types,  since  all  fell  within  the 
leptosome  range.  As  a  result,  only  the  correlation  tech- 
nique could  be  employed.1  As  in  the  study  on  men,  the 
correlations  between  test  scores  and  physical  indices  were 
too  low  to  be  of  any  significance. 

A  very  thojpiagLiliye^tig^tion  of  personality  traits  in 
relation  to  physical  type  has  recently  been  conducted  J^y 
KlmeB^  Eoky  (n).2  The  subjects  were 

again  students,  selected  from  several  colleges  in  New 
York  City  and  its  environs.  A  total  of  200  men  and 
229  women  were  examined.  Within  each  of  these  groups, 
the  subjects  who  fell  in  the  upper  and  those  who  fell  in 
the  lower  25%  of  the  distribution  of  Pignet  Index  were 
selected  as  leptosomes  and  pyknics,  respectively.  This 
gave  50  leptosomes  and  50  pyknics  among  the  men,  and 
57  leptosomes  and  57  pyknics  among  the  women. 

That  the  groups  so  obtained  were  clearly  differentiated 


measures  and  indices  fou^dior^allth^subjects.  In  both 
male  and  female  groups,  the  differences  between  lepto- 
somes and  pyknics  were  statistically  reliable  in  all  physical 

1As  a  subsidiary  analysis,  the  subjects  were  classified  into  "leptoid"  and 
"less  leptoid, "  and  average  scores  on  these  two  groups  were  computed.  This 
procedure  could  not,  however,  be  expected  to  yield  very  significant  results. 

2  The  writer  is  indebted  to  Drs.  Klineberg,  Fjeld,  and  Foley  for  making  avail- 
able to  her  the  unpublished  data  of  this  study. 


measures  except  standing  and  sitting  height  and  the 
ratio  between  these  two.  Not  only  were  the  differences 
between  averages  very  large  and  reliable  in  all  other 
physical  measures  or  indices,  but  the  ranges  showed  little 
or  no  overlapping.  In  regard  to  age,  the  male  pyknics 
proved  to  be  slightly  older  on  the  average  than  the  lepto- 
somes, the  average  age  of  the  male  leptosomes  being 
20.17  years  and  that  of  the  male  pyknics  21.08  years. 
In  addition  to  being  very  slight,  this  age  difference  is 
such  as  to  exaggerate  any  of  the  alleged  psychological 
differences  between  leptosomes  and  pyknics.  Hence  such 
an  age  discrepancy  loads  the  dice  slightly  in  favor  of 
Kretschmer's  hypothesis  and  would  make  negative  find- 
ings all  the  more  conclusive.  Among  the  females,  the 
age  difference  was  negligible,  the  leptosomes  averaging 
19.73  and  the  pyknics  19.23  years. 

All  subjects  were  given  the  Bernreuter  Personality  In- 
ventory, This  test  was  scored  with  six  keys,  yielding 
measures  of  as  many  different  though  more  or  less  inter- 
related aspects  of  personality,  viz.,  neuroticism,  self- 
sufficiency,  introversion,  dominance,  self-confidence,  and 
sociability.1  The  Allport-Fernon  Study  of  Values  was 
also  administered.  This  test  is  designed  to  measure  the 
relative  prominence  of  six  basic  interests  or  motives  in 
personality,  the  theoretical,  economic,  aesthetic,  social, 
political,  and  religious.  As  will  be  noted,  this  classifica- 
tion is  based  on  Spranger's  proposed  list  of  evaluative 
attitudes  (cf.  above).  All  subjects  were  further  given  a 
specially  devised  test  of  suggestibility.  In  addition,  it  was 

1  Statistical  analysis  has  recently  shown  that  the  Bernreuter  test  can  be  re- 
garded as  measuring  two  fundamental,  independent  traits  which  have  been 
labeled  self-confidence  and  sociability  (cf.  Flanagan,  J.  C.  Factor  Analysis  in 
the  Study  of  Personality.  Stanford  Univ.,  Calif.:  Stanford  Univ.  Press,  1935. 
Pp.  103).  The  other  four  scores,  although  not  mutually  independent,  were  in- 
cluded in  this  study  since  they  correlate  highly  with  scores  on  other  current 
personality  tests  and  correspond  to  more  familiar  psychological  terminology. 


possible  to  administer  two  other  tests  to  many,  but  not 
all,  individuals.  One  of  these  was  an  honesty  test,  showing 
the  number  of  times  the  subject  cheated  on  what  seemed 
to  be  an  information  test  (Mailer  Test  of  Sports  and 
Hobbies).  The  other  was  a  specially  constructed  per- 
sistence test  which  measured  the  length  of  time  the  indi- 
vidual worked  on  an  insoluble  finger  maze  before  giving  up. 

In  Table  XII  will  be  found  the  average  scores  of  both 
male  and  female  leptosomes  and  pyknics  on  each  test. 
Critical  ratios  are  also  given  to  show  the  reliability  of  the 
differences  between  the  averages.  It  will  be  recalled  that 
if  the  critical  ratio  has  a  value  of  3  or  more,  the  difference 
is  regarded  as  statistically  significant.  In  interpreting 
the  data  of  Table  XII,  it  should  be  borne  in  mind  that 
on  the  Bernreuter  test,  the  positive  scores  indicate  a 
greater  degree  of  the  trait  named,  the  negative  scores  a 
lesser  degree.  Thus  in  neuroticism,  for  example,  a  higher 
positive  score  indicates  a  more  neurotic  individual,  a  higher 
negative  score  a  less  neurotic  individual.  In  introversion, 
the  positive  scores  indicate  a  tendency  toward  introversion 
and  the  negative  scores  a  tendency  toward  extroversion. 
The  same  is  true  of  the  other  traits;  the  end  of  the  scale 
corresponding  to  the  positive  scores  is  the  one  from 
which  each  test  was  named.  In  the  Allport-Vernon  test, 
the  higher  scores  indicate  greater  prominence  of  the  partic- 
ular "value"  in  question.  In  suggestibility,  the  higher 
scores  show  a  greater  susceptibility  to  suggestion.  In  the 
persistence  test,  likewise,  the  higher  scores  show  greater 
persistence.  On  the  honesty  test,  the  scores  were  expressed 
in  such  a  form  that  the  higher  values  corresponded  to  the 
more  "honest"  individual  who  cheats  less  often. 

The^data  of  Table  XII  are  clearly  negative  as  regards 

***i^^^^r^ttK^^i,fW^i^*.«,Y.i  ««,  •  -,      "  -,  *»«*  "  "'".»>*«fc"  «**i»Ww»»m.*w,  ?,.,,,  %,  ° 

type  theories!  "jNoiie'bi  th£jej^^ 

iiT'eitKer  niale  or  fcmeoig.srd.iable.    In  the  male 


groups,  the  critical  ratios  range  from  0.005  to  2.22;  in 
the  female  groups,  the  range  is  from  0.09  to  2.22.  None 
of  the  differences  is  three  or  more  times  as  large  as  its 
standard  error.  All  the  obtained  differences  could  there- 
fore have  resulted  from  chance  errors  of  sampling.  In 
each  sex  group,  there  is  only  one  critical  ratio  which 
reaches  or  exceeds  a  value  of  2,  and  this  occurs  on  a 
different*  test  in  the  two  cases.  In  the  male  group,  the 
largest  critical  ratio  was  obtained  on  the  honesty  test, 
which  gave  a  critical  ratio  of  only  0.25  in  the  female  group. 
Similarly,  the  largest  critical  ratio  in  the  female  group  is 
found  with  the  "  social  values"  score  on  the  Allport- 
Vernon  scale  and  this  gives  a  critical  ratio  of  only  0.39 
in  the  male  group. 

It  might  also  be  pointed  out  that  in  several  comparisons 
the  differences  between  leptosomes  and  pyknics  are  con- 
trary to  expectation.  For  example,  the  male  leptosome 
group  appears  more  "sociable"  on  the  Bernreuter  and 
seems  to  have  a  higher  sense  of  "social  value"  according 
to  the  Allport-Vernon  scale,  than  does  the  male  pyknic 
group.  The  average  scores  of  all  the  groups,  furthermore, 
fall  very  close  to  the  norms  for  college  men  and  women 
in  general.  Finally,  the  ranges  of  the  leptosome  and  pyknic 
groups  were  nearly  identical,  showing  an  almost  complete 
overlapping  of  distributions  on  all  personality  tests. 

To  be  perfectly  cautious,  one  might  conclude  that  the 
groblem  of  constitutional  types  remains  an  open  question. 
All  the  Setter  controlled  studies  by  any  of  the  three 

<1-«^«-*«,^aJ*.,fa  „,»-„,   U,,!'..,^^     ,        ,,,<        ,       ...         '         ,•    ,      •  '  '  "  J  <*'        '  '•      ,  ,       '""•      --'«*«» 

nigthods  described  afepy^  ,.  have^  however,  ~  yielded^  ,o  ver- 

In  those  studies  which 

have  reported  positive  results,  it  is  a  relatively  easy 
matter  to  find  uncontrolled  factors  which  could,  in  them- 
selves, have  produced  the  observed  differences  in  alleged 
typal  characteristics.  Since  the  concept  of  types  seems  to 




(After  Klineberg,  Fjeld,  and  Foley,  n) 





(N  =  So) 


Ratio  of 



Ratio  of 


r.  BiNiNeuroti- 








2.  B2S:  Self-suffi- 



+  29.22 


—  0.26 

+  18.02 


3.  B3I:  Introversion 





—  27.00 


4.  B4D:  Dominance 
5.  Fid  Self-confi- 








-  B.S4 

-  8.46 


+  4.67 



6.  F2S:  Sociability 

+    1.22 

-  6.66 


—  29.79 




Study  of  Values 

I.  Theoretical 







2.  Economic 







3.  Esthetic 







4.  Social 







5.  Political 



i.  08 




6.  Religious 














Persistence  * 







Honesty  * 







*  Not  all  subjects  were' given  these  tests. 

have  proved  unsuccessful  in  explaining  the  organization 
of  traits  within  the  individual,  we  may  now  inquire  into 
the  latter  question  from  a  more  empirical  viewpoint. 
An  analysis  of  variations  within  the  individual,  to  be 
discussed  in  the  following  chapter,  will  serve  as  a  back- 
ground for  this  problem. 



1.  Burchard,  E.  M.  L.    "Physique  and  Psychosis:  an  Analysis 
of  the  Postulated  Relationship  between  Bodily  Constitution 
and   Mental   Disease    Syndrome,"    Comp.   Psychol.   Mon., 
1936,  13,  No.  i,  Pp.73- 

2.  Davenport,   C.   B.      "  Body-Build   and   Its   Inheritance," 
Carnegie  lust.  Wash.  Publ.,  No.  329,  1923.   Pp.  176. 

3.  Galton,  F.    Inquiries  into  Human  Faculty  and  Its  Develop- 
ment. London:  Macmillan,  1883.   Pp.  387. 

4.  Garrett,  H.  E.,  and  Kellogg,  W.  N.     "The  Relation  of 
Physical  Constitution  to  General  Intelligence,  Social  In- 
telligence, and  Emotional  Stability,"  /.  Exper.  Psychol., 
1928,  u,  113-129. 

5.  Garvey,    C.   R.      "Comparative   Body   Build   of  Manic- 
Depressive   and   Schizophrenic   Patients,"    Psychol.   £ulL, 
*933,  3°,  567-568  (see  also  p.  739). 

6.  Heidbreder,  E.    "Intelligence  and  the  Height- Weight  Ra- 
tio," /.  Appl.  Psychol.,  1926,  10,  52-62. 

7.  Hull,  C.  L.  Aptitude  Testing.  N.  Y.:  World  Book  Co.,  1928. 
Pp.  535- 

8.  James,  W.     Pragmatism,  a  New  Name  for  Some  Old  Ways 
of  Thinking.  N.  Y.:  Longmans,  1907.  Pp.  309. 

9.  Jung,  C.  G.    Psychological  Types  (transl.  by  H.  G.  Baynes). 
N.  Y.:  Harcourt,  Brace,  1924.  Pp.  654. 

10.  Klineberg,  0.,  Asch,  S.  E.,  and  Block,  H.  "An  Experimental 
Study  of  Constitutional  Types,"  Genet.  Psychol.  Hon.,  1934, 
1 6,  140-221. 

11.  Klineberg,  0.,  Fjeld,  H.?  and  Foley,  J.  P.,  Jr.   "An  Experi- 
mental Study  of  Personality  Differences  among  Constitu- 
tional, *  Racial,3  and  Cultural  Groups."    (To  appear.) 

12.  Kliiver,  H.    "An  Analysis  of  Recent  Work  on  the  Problem 
of  Psychological  Types,"  /.  Nerv.   and  Mental  Diseases, 
1925,  62,  561-596. 

13-  •   "The  Problem  of  Type  in  "Cultural  Science  Psychol- 
ogy/ "  /•  Phttos.,  1925,  22,  225-234. 

14-  •    "Studies  on  the  Eidetic  Type  and  on  Eidetic  Im- 
agery," Psychol.  Bull,  1928,  25,  69-104. 


15.  Kretschmer,  E.    Physique  and  Character  (transl.  from  2nd 
ed.  by  W.  J.  H.  Sprott).     N.  Y.:  Harcourt,  Brace,  1925. 
Pp.  266. 

1 6.  .    Korperbau  und  Charakter  (Tenth  edition).     Berlin: 

Springer,  1931.  Pp.  240. 

17.  McDougall,  W.     Outline  of  Abnormal  Psychology.    N.  Y.: 
Scribner's,  1926.  Pp.  572. 

1 8.  Mohr,  G.  H.,  and  Gundlach,  R.  H.   "The  Relation  between 
Physique  and  Performance,"  /.  Exper.  PsychoL,  1927,  10, 

19.  Naccarati,  S.    "The  Morphologic  Aspect  of  Intelligence," 
Arch.  PsychoL,  No.  45,  1921.   Pp.  44, 

20.  .     "The  Morphologic  Basis  of  the  Psychoneuroses," 

Amer.  J.  Psychiat.,  1924,  3,  527-545. 

21.  Naccarati,  S.,  and  Garrett,  H.  E.    "The  Relation  of  Mor- 
phology to  Temperament,"  /.  Abn.  and  Soc.  PsychoL,  1924- 
*5,  *9,  254-263. 

22.  Pavlov,  I.  P.     Conditioned  Reflexes:  an  Investigation  of  the 
Physiological  Activity  of  the  Cerebral  Cortex  (transl.  and  ed. 
by  G.  V.  Anrep).    Oxford  Univ.  Press:  Humphrey  Milford, 
1927.   Pp.430. 

23.  Sheldon,  W.  H.   "Morphologic  Types  and  Mental  Ability," 
/.  Pers.  Res.,  1927,  5,  447-4S1- 

24.  .     "Social  Traits  and  Morphologic  Types,"  /.  Pers. 

Res.,  1927,  6,  47-55. 

25.  Spranger,  E.     Types  of  Men  (transl.  by  P.  J.  W.  Pigors). 
Halle:  Niemeyer,  1928.  Pp.  402. 

26.  Stockard,  C.  R.    "Human  Types  and  Growth  Reactions," 
Amer.  J.  Anat.,  1923,  31,  261-288. 

27.  Weidenreich,  F.     Rasse  und  Korperbau.     Berlin:  Springer, 
1927.  Pp.  187. 

28.  Wertheimer,  F.  L,  and  Hesketh,  F.  E.    "The  Significance 
of  the  Physical  Constitution  in  Mental  Disease,"  Medicine, 
1926,  5  ^  37S~463- 


f  variations  from  trait  to  trait  within  the 
individual  is  of  both  practical  importance  and  theoretical 
significance.  When  a  child  is  classified  as  intellectually* 
inferior  on  the  basis  of,  let  us  say,  Stanford-Binet  I.Q., 
there  is  still  much  that  remains  to  be  known  about  his 
mentality.  Is  he  equally  inferior  in  all  respects  or  does  he 
exhibit  significant  discrepancies  in  his  mental  develop- 
ment? Is  he  normal  or  even  superior  along  some  specific 
lines  ?  Similarly,  in  the  case  of  a  child  of  very  high  I.Q., 
we  may  inquire  in  what  ways  he  is  superior.  How  uni- 
formly does  he  excel  the  average  child  in  intellectual  per- 
formance? The  intelligence  test,  furnishing  a  single  sum- 
mary -  figure  to  characterize  the  child's  general  mental 
level,  often  obscures  important  facts.  Two  individuals 
obtaining  the  same  total  score  may  present  very  different 
"mental  pictures"  when  their  performance  along  specific 
lines  is  analyzed. 

The  skilled  clinician  or  mental  tester  has  always  taken 
this  into  consideration  in  interpreting  test  scores.  The 
child's  performance  on  the  different  parts  of  an  intelligence 
scale  and  even,  when  feasible,  on  several  different  kinds  of 
tests,  is  carefully  analyzed  before  a  final  judgment  is 
offered.  In  this  way,  the  practical  common  sense  of  the 
examiner  is  brought  in  to  remedy  a  deficiency  in  the  test 
itself.  There  is  a  rapidly  growing  realization,  however, 
that  the  question  of  variation  among  the  individual's 
abilities  deserves  serious  and  systematic  consideration  and 
should  be  investigated  in  its  own  right.  This  problem  is 



gradually  coming  to  be  regarded  as  even  more  important 
than  the  establishment  of  the  individual's  general  level  of 

In  planning  an  educational  program  for  a  given  in- 
dividual, or  in  helping  him  to  choose  a  vocation,  it  is  of 
the  greatest  importance  to  know  his  strong  and  his  weak 
points.  Total  scores  on  intelligence  tests  can  be  used  only 
inva  very  crude""affd  general  sort  of  educational  and  vo- 
cational guidance.  In  the  comparative  study  of  groups, 
such  as  the  sexes  or  different  racial  or  cultural  groups,  a 
consideration  of  the  general  level  of  ability  may  also  prove 
misleading.  Let^u&.^uppose,. .  for  example, -that-  one^such 
group  excels  markedly  in  ability  A  and  the  other  m  abjlity 
Bs.... If  both  are  examined,  with  an  intelligence  test  jvhich 
55J5P-!e?  Cities  A  and, B,, equal  extent,  no  difference 
*n  Jfital  scoi*e  will  appear  jDetweea  rthje  ,twp  groups.  Essen- 
tial and  large  differences  might  thus  be  concealed  by  the 
practice  of  lumping  a  number  of  tasks  indiscriminately  in 
the  effort  to  arrive  at  the  general  mental  average  called 

Much  confusion  has  likewise  been  introduced  into  the 
interpretation  of  test  results  by  the  common  tendency  to 
take  labels  too  seriously.  Thus,  if  two  tests  are  labeled 
measures  of  "  intelligence,"  it  is  assumed  that  they  are 
measuring  the  same  characteristic  of  the  individual.  It  is 
therefore  most  disconcerting  to  discover  that  the  same 
child  may  appear  dull  in  one  intelligence  test  and  above 
average  in  another.  Such  cases  are,  however,  found. 
Since  intelligence  scales  consist  of  a  more  or  less  random 
sampling  of  different  tasks,  the  specific  abilities  covered 
by  the  various  tests  may  differ.  Some  tests,  for  example, 
may  be  more  heavily  "loaded"  with  verbal  items,  others 
with  mechanical  items.  -Even  successive  levels  of  the  same 
test  occasionally  involve  different  abilities.  Thus  the 


Stanford-Binet  draws  more  heavily  from  the  verbal  field 
at  the  higher  year  levels  than  it  does  at  the  lower.  The 
same  child  tested  with  the  Stanford-Binet  at  different 
ages  might  be  favored  at  one  time  and  handicapped  at 
another  because  of  the  particular  abilities  called  into  play. 
If  the  individual's  abilities  were  all  more  or  less  on  a 
dead  level,  a  single  summary  score  would  be  quite  informa- 
tive. But  if  appreciable  variation  in  the  individual's 
standing  in  different  traits  is  the  rule,  then  such  a  score 
is  crude  at  best  and  may  upon  occasion  be  definitely  mis- 
leading. It  is  essential,  therefore,  to  inquire  into  the  extent 
of  variation  within  the  individual.  The  data  on  this 
question  have  been  gathered  from  a  variety  of  sources. 
Case  studies  are  available  of  individuals  who  exhibit 
marked  asymmetry  of  development  along  different  lines. 
Such  individuals  can  be  found  among  the  feebleminded 
and  the  intellectually  superior,  as  well  as  among  the  nor- 
mal. Quantitative  measurements  have  also  been  made  on 
the  extent  of  variability  from  trait  to  trait  in  large  random 
samplings.  Finally,  correlational  analysis  throws  some 
light  upon  this  problem.  Typical  data  obtained  by  these 
various  approaches  will  be  examined. 


Among  the  feebleminded,  persons  are  occasionally  met 
who  display  an  exceptional  talent  along  some  specific  line. 
Such  individuals  have  been  designated  "idiots  savants" 
(wise  idiots).  This  term  has  been  criticized  for  being 
somewhat  misleading.  The  usual  idiot  savant  is  neither 
particularly  wise  nor  an  idiot.  He  is  not  sufficiently  defi- 
cient to  be  classified  as  an  idiot,  but  is  frequently  found  at 
the  moron  or  borderline  level.  And  he  is  "wise"  only  in  a 
very  limited  field.  In  the  practical  management  of  his 
own  life  he  is  ordinarily  a  complete  failure.  Many  of  the 


idiots  savants  in  institutions  for  the  feebleminded  were 
physically  awkward  or  uncouth,  had  a  serious  sensory 
deficiency,  or  were  reputed  to  be  emotionally  unstable  or 
irresponsible.  Otherwise  they  might  have  been  able  to 
shift  for  themselves  with  a  little  assistance,  by  capitalizing 
their  special  talent.  As  is  true  of  all  extreme  deviations  in 
the  distribution  of  any  trait,  idiots  savants  are  relatively 
rare.  Because  of  their  striking  quality,  however,  such 
cases  attract  a  good  deal  of  attention,  so  that  a  number  of 
fairly  complete  descriptive  accounts  are  now  available. 

ThejpeciaHalent  of  the  idiot  savant  maj^  be  observed  in 
j>£a<^^  activity.     Mechanical 

aptitude,  ability  in  drawing  and  painting,  a  phenomenal 
memory,  arithmetic  proficiency,  a  special  gift  in  music,  all 
are  represented.  Th£jon£j5eld  from  which  idiots  savants 
sf£SL?*L k?,  conspicuously . , absentJs , that'  Q|  linguistic  or 
verbal  aptitude.1  This  fact  throws  some  light  upon  our 
concept  ol  "general  intelligence.  As  will  be  more  fully 
demonstrated  in  the  following  chapter,  the  latter  is  very 
largely  identified  with  verbal  ability.  Most  intelligence 
tests  are  composed,  to  a  very  great  extent,  of  verbal  tasks. 
Success  in  the  practical  business  of  everyday  life  is  also 
more  closely  linked  to  linguistic  facility  than  to  other 
traits.  A  serious  deficiency  in  the  power  of  verbal  expres- 
sion will  thus  brand  an  individual  as  incompetent  from 
many  points  of  view.  Conversely,  a  person  who  is  es- 
pecially proficient  in  verbal  traits  may  thereby  compensate 
for  deficiencies  along  other  lines  and  will  rarely,  if  ever, 
find  his  way  into  an  institution  for  the  feebleminded.  No 

1  Feebleminded  individuals  have  been  described  who  could  repeat  long  quota- 
tions in  several  foreign  languages,  without  error  and  with  perfect  accent.  But 
these  cases  probably  represent  a  more  specific  abnormal  phenomenon  and  can- 
not be  regarded  as  having  superior  linguistic  endowment  in  the  usual  sense. 
Such  individuals,  for  example,  are  frequently  unable  to  read  or  write.  (Cf. 
Barr,  4,  for  a  report  of  such  a  case.) 


other  single  talent  seems  to  be  such  a  saving  grace  in  our 

Several  cases  of  special  talent  in  pictorial  art  have  been 
found  among  the  feebleminded.  Such  individuals  are  able 
to  execute  excellent  reproductions  of  well  known  paintings. 
Occasionally  this  talent  passes  beyond  mere  copying  and 
suggests  real  creative  genius.  Such  a  case  is  that  of  Gott- 
fried Mind  (35,  p.  336)3  diagnosed  as  a  cretin  imbecile. 
His  mental  deficiency  was  manifested  from  an  early  age. 
He  was  unable  to  learn  to  read  and  write.  His  movements 
were  awkward,  his  hands  large  and  rough,  and  his  general 
appearance  that  of  the  traditional  mental  defective.  Since 
he  showed  considerable  talent  for  drawing,  he  was  given 
some  instruction  in  this  field.  His  subsequent  success  in 
pictorial  art  was  phenomenal.  Because  of  his  excellent 
drawings  of  cats,  he  came  to  be  known  as  aThe  Cats' 
Raphael"  In  addition,  he  produced  drawings  and  water- 
color  sketches  of  deer,  rabbits,  bears,  and  groups  of  chil- 
dren, which  were  remarkable  for  their  life-like  quality  and 
masterly  execution.  His  fame  spread  throughout  Europe 
and  one  of  his  pictures  of  a  cat  and  kittens  was  purchased 
by  King  George  IV  of  England. 

An  equally  remarkable  case  is  that  of  J.  H.  Pullen,  who 
has  been  called  "The  Genius  of  Earlswood  Asylum"  (35, 
pp.  340-345).  This  individual  had  extraordinary  mechan- 
ical ingenuity  coupled  with  talent  in  drawing  and  carving. 
In  other  respects  he  was  very  deficient.  He  did  not  talk 
until  the  age  of  seven,  and  for  a  long  time  only  uttered 
the  word  "muvver."  His  speech  was  always  imperfect, 
and  he  is  reported  as  being  very  deaf.  He  was  taught  by 
his  family  to  write  and  to  spell  the  names  of  simple  objects, 
and  this  constituted  the  extent  of  his  schooling.  From  an- 
early  age,  he  spent  much  of  his  time  in  drawing  or  in  carv- 
ing ships  out  of  pieces  of  firewood,  occupations  in  which  he 


showed  considerable  proficiency.  At  the  age  of  15,  he  was 
admitted  to  Earlswood  Asylum,  where  he  was  put  to  work 
in  the  carpenter's  shop  and  soon  became  an  expert  crafts- 
man. During  his  sixty-six  years  at  the  asylum,  he  pro- 
duced an  impressive  array  of  beautiful  and  highly  in- 
genious objects,  including  crayon  drawings,  carvings  in 
ivory  and  wood,  excellent  models  of  ships,  and  various 
mechanical  devices.  Occasionally  he  even  designed  his 
own  instruments  to  help  him  in  his  work. 

One  of  Pullen's  constructions  was  a  representation  of 
a  gigantic  human  form,  thirteen  feet  high.  This  full- 
fashioned  "robot"  could  be  made  to  execute  a  variety 
of  movements  such  as  raising  the  arms,  rotating  the  head, 
protruding  the  tongue,  opening  and  shutting  mouth  or 
eyes,  etc.  Another  remarkable  construction  was  a  model 
of  a  ship,  beautifully  executed  in  the  minutest  detail. 
This  model  required  over  three  years  for  its  completion 
and  attracted  universal  admiration  when  exhibited.  Pul- 
len's work  revealed  artistic  imagination  as  well  as  mechan- 
ical ingenuity,  skill  in  planning,  and  painstaking  execution. 
Being  cut  off  from  many  ordinary  sources  of  stimulation 
by  deafness,  it  is  probable  that  he  concentrated  all  his 
efforts  from  childhood  upon  the  development  of  this  one 
remarkable  talent.  In  regard  to  general  personality  de- 
velopment, he  is  described  as  childish  and  immature,  emo- 
tionally unstable,  and  lacking  in  common  sense. 

Special  talent  in  music  has  also  been  repeatedly  ob- 
served among  the  feebleminded.  A  very  striking  case 
(35>  P-  339)  of  exceptional  musical  ability  combined  with 
serious  defect  in  other  respects  is  that  of  a  woman  in  the 
Saltpetriere,  a  famous  French  institution  for  the  feeble- 
minded and  the  insane.  This  patient  was  an  imbecile, 
blind  from  birth,  a  cripple,  and  affected  with  rickets. 
She  was,  however,  able  to  sing  without  error  any  selection 


which  she  had  heard.  It  became  customary  for  her  fellow- 
inmates  to  come  to  her  so  that  she  might  correct  their 
mistakes  in  singing.  She  attracted  wide  attention,  and  one 
day  the  composers  Liszt  and  Meyerbeer  visited  her  "  sing- 
ing class "  to  bring  their  encouragement  and  consolation. 

Recently,  a  similar  example  of  musical  talent  has  been 
investigated  by  the  use  of  standardized  intelligence  tests 
(cf.  26).  This  was  the  case  of  a  boy  admitted  to  a  feeble- 
minded institution  at  the  age  of  14.  He  came  from  an 
intellectually  superior  family  which  included  many  musi- 
cally gifted  individuals  among  its  members.  As  a  child, 
the  subject  was  intellectually  normal  and  manifested  his 
musical  talent  from  an  early  age.  When  three  years  old, 
he  had  pneumonia  and  meningitis,  and  since  then  he  under- 
went steady  mental  deterioration.  Upon  admission,  his 
I.Q.  was  62;  at  the  last  testing,  it  had  dropped  to  46. 
He  was  then  over  20  years  of  age  and  had  a  mental  age  of 
7  years-5  months.  His  memory  was  unimpaired,  how- 
ever, and  he  retained  his  excellent  musical  ability.  Al- 
though never  known  to  compose  a  piece,  he  could  play 
difficult  music  by  ear  and  was  also  able  to  read  difficult 
musical  compositions  at  sight. 

The  feats  of  memory  performed  by  some  feebleminded 
individuals  have  often  attracted  notice.  Tredgold  (35, 
p.  337)3  for  example,  describes  a  65-year-old  mental  de- 
fective in  Earlswood  Asylum  with  a  remarkable  memory 
for  historical  facts.  He  could  repeat  the  dates  of  birth 
and  death  and  the  essential  facts  of  the  life  of  any  prom- 
inent character  in  history.  This  knowledge  was  acquired 
largely  by  rote,  through  poring  over  all  available  books  on 
biography  and  history.  It  was  not,  however,  a  matter  of 
sheer  meaningless  repetition,  as  was  shown  by  the  subject's 
responses  when  questioned  on  the  material.  Another  pa- 
tient at  the  same  institution  showed  an  excellent  memory 


for  dates  and  occurrences  which  had  come  within  his  own 
experience.  He  proved  a  useful  source  of  reference  on  local 
happenings  in  the  institution. 

Arithmetical  prodigies  have  also  been  found  among  the 
ranks  of  the  feebleminded.  Usually,  the  skill  manifested 
is  confined  to  the  mechanics  of  computation.  Thus  the 
subject  may  perform  long  and  complicated  calculations 
within  a  very  short  time  and  without  the  aid  of  paper  and 
pencil.  A  favorite  feat  is  to  determine  the  number  of  min- 
utes a  person  has  lived,  from  a  knowledge  of  his  age  or 
date  of  birth.  Multiplication  of  three-place  numbers,  nam- 
ing square  roots  and  cube  roots  of  four-place  numbers,  and 
similar  difficult  operations  have  also  been  executed  within 
a  few  seconds.  In  some  cases,  this  numerical  aptitude  goes 
beyond  routine  computation,  as  is  indicated  by  the  in- 
dividual's ability  to  solve  mathematical  problems  ex- 
pressed in  fairly  elaborate  and  confusing  terms. 


Asymmetry  of  mental  development  is  not  to  be  regarded 
as  characteristic  of  the  feebleminded.  It  is  equally  com- 
mon outside  of  institutions  and  among  those  classified  as 
normal  or  superior  on  the  basis  of  intelligence  test  per- 
formance. As  was  found  in  the  case  of  the  feebleminded, 
verbal  traits  are  closely  linked  with  what  is  termed  " gen- 
eral intelligence"  and  therefore  offer  no  examples  of  special 
talent.  Children  who  are  deficient  in  reading  or  spelling 
are  usually  inferior  on  intelligence  tests.1  Occasionally,  a 
young,  bright  child  is  a  poor  speller  because  of  his  distaste 
for  the  routine  drill  necessary  to  master  this  school  sub- 
ject, but  the  inferiority  is  usually  overcome.  Juvenile 
authors,  furthermore,  have  invariably  been  children  of 
very  high  LQ.2  In  many  other  traits,  however,  marked 

1  Cf.  L.  S.  Hollingworth.  (16,  Ch.  IV  and  V).     2  Cf.  17,  Ch.  IX,  for  examples. 

266       '          DIFFERENTIAL  PSYCHOLOGY 

discrepancies  have  been  found  between  the  child's  alleged 
"general  mental  level"  and  his  ability  along  specific  lines. 
Musical  aptitude  seems  to  have  little  or  no  relationship 
to  ST^eno^ntelligence.  This  was  clearly  shown  in  a  study 
by*L.  S.  Holiingworth  (18)  qn^g  intellectually  gifted 
cjiildren.  All  of  the  subjects  were"' enrolled  in  special 
classes  conducted  for  children  with  LQ.'s  of  135  or  higher. 
The  median  Stanford-Binet  I.Q.  of  this  group  was  153 
and  the  range  extended  from  135  to  190.  la  chronological 
age,jthe  children  corresponded, closely  to  thet,group  of  fifth 
gradeschoolchildren  upon  whom  the  Seashore  tests  of 
musical  sensitivity  l  had  been  standardized.  Accordingly, 
the  scores  of  the  intellectually  superior  children  on  these 
tests  were  expressed  as  percentiles  of  Seashore's  fifth  grade 
group.  The  average  percentile  scores  thus  obtained  on 
each  test  are  given  below. 

Test  Average  Percentile  Score 

Pitch  47 

Intensity  50 

Time  58 

Consonance  48 

Tonal  memory  52 

A  percentile  score  of  50,  it  will  be  recalled,  corresponds  to 
the  middlemost  score  of  the  standardization  group  and 
thus  represents  a  "normal/'  or  average,  performance. 
The  closeness  to  50  of  all  the  average  percentile  scores  of 
the  superior  group  indicates  that  musical  aptitude  is  dis- 
tributed among  these  children  in  very  much  the  same 
fashion  as  in  any  group  of  the  same  age,  chosen  at  random. 
Although  in  intelligence  test  performance  these  subjects 
were  all  within  the  upper  i%  of  the  general  population, 

1  For  a  description  of  these  tests,  cf.  Seashore,  C.  E.   The  Psychology  of  Musi- 
cal Talent.  N.Y.: Silver,  Burdett,  1919.  Pp.288. 


their  individual  percentile  scores  on  the  music  tests  ranged 
from  zero  to  98. 

Case  studies  of  arithmetical  prodigies  and  "  lightning 
calculators''  suggest  that  numerical  aptitude  is  likewise 
distributed  independently  of  general  intelligence.  Many 
such  cases,  from  the  early  Greeks  to  the  end  of  the  last 
century,  have  been  brought  together  by  Scripture  (31) 
and  later  by  Mitchell  (27).  In  r^ai^to^lieir%;achievements 
along  other  lines^or  their  practical  ability  to  sucqeecfin 
every^ay^Hfe^mathematicaLpiociigies  rua  the  gamut,  from 
genius  and  eminence  of  the  highest  order  to  mental,  dull- 
ness."  A  few  would  no  doubt  be  classified  as  "borderline" 
or  lower  on  current  intelligence  tests.  At  the  other  ex- 
treme are  such  men  as  Gauss  and  Ampere,  whose  excep- 
tional talents  covered  a  wide  range,  and  who  have  made 
distinguished  contributions  in  mathematics  and  allied 
fields.  These  men  were  "  lightning  calculators,"  but  also 
possessed  very  superior  ability  along  many  other  lines. 
For  the  present  purpose,  however,  we  are  concerned  with 
cases  of  asymmetrical  development  in  which  prodigious 
arithmetic  powers  are  coupled  with  mediocrity  or  defi- 
ciency in  other  respects. 

Henri  Mondeaux  (cf.  31),  the  untutored  son  of  a  poor 
woodcutter,  is  a  famous  example  of  remarkable  arithmetic 
ability  in  an  otherwise  dull  person.  In  his  childhood  he 
received  no  instruction,  but  was  sent  to  tend  sheep  at  the 
age  of  12.  While  engaged  in  this  occupation,  he  amused 
himself  by  counting  and  arranging  pebbles;  by  this  means 
he  learned  to  carry  out  arithmetic  operations.  He  worked 
out  for  his  own  use  many  special  devices  and  aids  to  com- 
putation. After  long  exercises  at  these  calculations,  he 
offered  to  tell  people  he  met  the  number  of  seconds  in 
their  ages.  At  this  time,  a  schoolmaster  became  interested 
in  him  and  offered  to  instruct  him.  Unfortunately  the 


boy  had  a  very  poor  memory  for  names  and  addresses 
and  he  spent  nearly  a  month  searching  the  city  before  he 
was  able  to  locate  his  benefactor.  Mondeaux  was  subse- 
quently exhibited  by  his  teacher  at  several  colleges  and 
universities  and  in  1840  he  was  presented  before  the 
Academy  of  Sciences  at  Paris.  His  was  not  merely  a  talent 
for  routine  calculation.  He  demonstrated  his  ability  to 
solve,  by  ingenious  devices  of  his  own  making,  complex 
problems  such  as  the  following: 

There  is  a  fountain  containing  an  unknown  quantity  of 
water;  around  it  stand  people  with  vessels  capable  of  con- 
taining a  certain  unknown  quantity.  They  draw  at  the  fol- 
lowing rate:  the  first  takes  loo  quarts  and  -fa  of  the  remainder; 
the  second  takes  200  quarts  and  TV  °f  the  remainder;  the 
third  300  quarts  and  TV  °f  'the  remainder,  and  so  on  until 
the  fountain  is  emptied. 

Mondeaux  gave  the  correct  answer  to  this  problem  in  a 
few  seconds  and  then  explained  the  method  whereby  he 
had  arrived  at  the  solution. 

A  similar  case  is  that  of  Tom  Fuller,  a  Negro  slave 
brought  from  Africa  at  the  age  of  14.  He  could  neither 
read  nor  write  and  received  no  formal  instruction.  As  in 
the  case  of  Mondeaux,  his  arithmetic  was  entirely  self- 
taught.  It  is  reported  of  him  that  when  asked  how  many 
seconds  a  man  had  lived  who  is  70  years-iy  days-12 
hours  old,  he  gave  the  answer,  after  ij  minutes,  as 
2,210,500,800  seconds.  One  of  his  questioners  had  mean- 
time been  computing  with  paper  and  pencil  and  informed 
Fuller  that  he  had  arrived  at  a  different  number,  which  he 
read  off.  At  this,  the  Negro  immediately  pointed  out  that 
his  questioner  had  forgotten  to  allow  for  leap  years ! 

A  few  cases  of  "  lightning  calculators  "  have  been  directly 
observed  and  investigated  by  psychologists.1 

1  Cf.  Binet  (6)  and  Lindley  and  Bryan  (23) . 


tiQnjjxus.uQbtained,  as  well  a.s_  the.  careful  analysis- of  avail- 
ably reports  on  arithmetic  prodigies,  has  brought  to  light 
certain  characteristics  of  these  individuals  which  may 
account  for  their  talent.  Injmost  cases,  the  individual  has 
worked  out  a  number  of  short-cuts  and  special  devices 
which  ..enable  him  to  compute  far  more  efficiently  than  is 
ordinarily  possible.  Secondly,  such  .individuals  have  usu- 
ally .memorized  many  more  number  combinations,  such  as 
squares,  cubes,  roots,  products,  etc.,  than  are  at  the  dis- 
posal of  the  average  man.  Arithmetical  prodigies  in- 
variably manifest  a  very  keen  interest  and  fascination 
for  numbers;  as  a  result,  they  devote  much  time  to  analysis 
of  computation  methods  and  drill  which  would  otherwise 
prove  very  monotonous.  Mairy^also  seem  to  have  a  large 
perception  3pan,  which  enables  them  to  grasp  a  long  series 
of-mimbers  simultaneously,  as  well  as  vivid  imagery,  mak- 
ing_pQ3sihle  " mental  computation"  without  the  aid?  of 
paper  and  pencil. 

The  examples  which  have  been  reported  suffice  to  illus- 
trate the  existence  of  extreme  asymmetries  of  mental  de- 
velopment. It  is  apparent  that  talent  along  certain  lines 
is  not  incompatible  with  mediocre  or  inferior  status  in 
others.  The  cases  so  far  described  represent  extreme  devi- 
ations which  for  that  reason  are  easily  recognizable.  Other 
instances  of  special  talents  or  deficiencies,  not  so  striking 
as  to  attract  widespread  attention,  but  revealed  by  the 
application  of  quantitative  techniques  of  measurement, 
will  be  found  in  the  following  section. 


In  the  effort  to  obtain  a  more  objective  and  concrete 
picture  of  variations  within  the  individual  than  is  furnished 
by  the  general  impression  of  the  examiner,  a  psychograph, 
or  profile  chart*  of  the  individual  may  be  drawn  up.  The 


psychograph  shows  at  a  glance  the  relative  standing  of  the 
subject  on  any  number  of  tests  or  other  measures.  The 
individual's  scores  on  all  tests  must  first  be  transmuted 
into  comparable  units.  This  is  the  fundamental  step  in  any 
attempt  to  study  variations  within  the  individual.  The 
psychograph  itself,  in  the  sense  of  a  pictorial  representa- 
tion, could  easily  be  dispensed  with.  The  same  informa- 
tion,, although  in  a  less  vivid  form,  could  be  got  from  an 
examination  of  a  set  of  scores  obtained  by  the  individual, 
provided  that  all  scores  are  expressed  in  the  same  terms.  It 
is  in  this  latter  respect  that  the  judgment  of  the  examiner 
needs  to  be  supplemented  by  quantitative  techniques. 
Confronted  with  a  set  of  scores,  some  of  which  are  ex- 
pressed in  seconds,  others  as  number  of  words  recalled,  and 
still  others  as  number  of  problems  correctly  solved,  the 
clinician  would  be  at  a  complete  loss. 

Comparable  measures  can  be  obtained  in  several  ways. 
If  all  the  tests  have  been  standardized  in  terms  of  age, 
each  score  can  be  expressed  as  a  ^en^l^ge.1  In  many 
situations,  however,  this  is  not  feasible.  Some  tests,  es- 
pecially in  the  field  of  personality,  do  not  exhibit  large  or 
systematic  age  changes.  The  range  of  variation  within 
one  age  group  might  thus  be  greater  than  the  largest  dif- 
ference between  the  average  performance  of  age  groups. 
The  application  of  the  mental  age  concept  to  adult  sub- 
jects, furthermore,  is  a  rather  questionable  practice. 

A  commonly  employed  and  generally  applicable  meas- 
ure is  the^^£n^^  score  (cf.  Ch.  II).  A  percentile,  it  will 
be  recalled,  gives  tET^percentage  of  individuals  whose 
scores  fall  at  or  below  that  obtained  by  the  given  subject. 
Percentiles  are  determined  once  and  for  all  in  the  process 

1 A  mental  age  psychograph  is  employed  at  the  N.  Y.  C  Children's  Hospital, 
an  institution  for  the  feebleminded.  (Cf.  Poull,  L.  E.  "  The  Psychographic 
Method  in  Clinical  Practice."  /.  Appl.  PsychoL,  1936,  20,  161-164.) 


of  standardization.  The  test  is  administered  to  a  large 
group,  representative  of  the  population  upon  which  it 
will  ultimately  be  employed,  and  the  percentage  of  in- 
dividuals who  score  at  or  below  each  point  is  determined. 
Thus,  if  on  an  arithmetic  reasoning  test,  10%  of  the  sub- 
jects correctly  solve  three  problems  or  less,  then  any  child 
who  completes  three  problems  correctly  receives  a  per- 
centile  score  of  10.  A  percentile  scale  divides  the  group 
into  100  classes,  each  composed  of  the  same  number  of 
persons.  Any  subject  subsequently  taking  the  test  is  then 
placed  into  one  of  these  classes,  the  poorest  corresponding 
to  the  first  percentile  and  the  best  to  the  looth  percentile. 
A  subject  who  receives  a  percentile  rating  of  100  has  ob- 
tained the  highest  score  reached  in  the  standardization 
group,  but  not  necessarily  the  maximum  score  possible  on 
the  test.  Similarly,  a  zero  percentile  rating  does  not  mean 
a  zero  score;  it  signifies  only  that  the  subject's  score  is 
lower  than  that  obtained  by  any  member  of  the  standard- 
ization group. 

Scores  from  different  tests  can  also  be  made  comparable 
by  the  use  of  standard  measures.1  In  this  case,  the  subject's 
score  is  exp res s edT^FT'Heviation  above  or  below  the  aver- 
age of  a  given  group.  Thus  if  his  original  score  falls  exactly 
at  the  average,  he  receives  a  standard  score  of  zero.  If  he 
is  better  than  average,  he  receives  a  plus  score,  if  poorer 
than  average,  a  minus  score.  The  unit  in  terms  of  which 
the  scores  are  reported  is  the  standard  deviation  (S.D.)  of 
the  distribution.  Thus,  if  the  average  of  the  distribution  is 
35  and  the  S.D.  10,  and  if  a  given  subject  obtains  a*  score  of 
45  points,  his  standard  score  would  be  +I3  or  one  standard 

/ A  r>   o  r  \ 

deviation  above  the  group  average  ( — — —  =  i  ).    Simi- 
larly, if  another  subject  receives  a  score  of  30  on  the  same 
1  Cf.  Kelley  (21),  pp.  114-117. 


(3°  ~~  35  A 

— — —  =  — .5!- 

If  the  distribution  of  scores  on  the  different  tests  can  be 
assumed  to  be  of  the  same  general  shape,  approximating 
the  normal  bell-shaped  curve,  the  standard  scores  com- 
puted from  them  will  be  directly  comparable.1 

It  should  be  borne  in  mind  that  none  of  these  techniques 
for  converting  scores  into  comparable  measures  yields  a 
scale  of  equal  units.  They  simply  express,  in  terms  which 
are  more  or  less  int^lVi^^^^^  relative  position  oi  the  in- 
dividual in  different  tests,  but  they  do  not  furnish  a  precise 
statement  of  the  actual  amount  of  trait  difference  repre- 
sented by  the  various  scores.  Thus,  it  will  be  recalled  that 
the  mental  age  unit  corresponds  to  the  average  change  in 
score  occurring  during  a  one-year  period.  Successive  men- 
tal ages  will  not,  therefore,  represent  equal  increments  of 
ability.  We  know  that  an  M.A.  of  6  indicates  a  higher 
standing  than  an  M.A.  of  5,  and  that  an  M.A.  of  10  in- 
dicates a  higher  standing  than  one  of  9,  but  we  cannot 
conclude  that  the  amount  of  difference  is  the  same  in  both 

Nor  can  percentile  scores  be  interpreted  as  equal  ability 
units.  As  was  shown  in  Chapter  II,  such  an  interpretation 
would  imply  that  the  distribution  of  the  trait  measured  is 
rectangular.  Since,  however,  traits  are  distributed  accord- 
ing to  the  normal  bell-shaped  curve,  individuals  will 
cluster  more  closely  at  the  center  of  the  distribution  and 
scatter  as  the  extremes  are  approached.  As  a  result,  a 
difference  of  one  percentile  point  at  the  extremes  corre- 

1  Hull  (19)  has  described  a  technique  for  obtaining  comparable  measures  by 
transmuting  the  original  scores  into  a  distribution  with  any  desired  average  and 
S.D.  This  procedure  is  based  upon  the  same  principle  as  standard  measures. 
Its  chief  advantages  lie  in  the  fact  that  the  scores  can  be  expressed  in  more 
familiar  terms  than  by  the  use  of  standard  measures,  and  that  negative  values, 
decimals,  etc.,  can  be  eliminated. 


spends  to  a  much  greater  difference  in  amount  of  the  trait 
than  does  a  difference  of  one  percentile  point  nearer  the 
center.  The  difference  between  an  individual  who  re- 
ceives a  percentile  rating  of  90  in  height  and  one  who  re- 
ceives a  percentile  rating  of  91  is  much  greater,  in  actual 
inches^  than  the  difference  between  two  individuals  re- 
ceiving percentile  ratings  of  50  and  51. 

Similarljr^§taiidard.acore,a,dD  not  represent  equal  units. 
BjLSuhte&cting  a  constant  (the  average)  and  dividing  by  a 
cQastan.£._(the  $.D.)>  we^o^jiot,  Baiter  T,the  scares  in,  any 
The  set  of  measures  is  simply  transmuted 

into  a  different  system  of  expression,  as  when  pounds  are 
changed  to  kilograms.  Rirtthe.staadard  scores  so  .obtained 

which  were  present 

^liSSS^-SSS8}??;??*  X 
In  addition  to  the  different  ways  of  expressing  com- 

parable scores,  there  are  a  variety  of  ways  in  which  the 
psychograph  itself  can  be  plotted.  Illustrations  of  differ- 
ent kinds  of  psychographs  will  be  found  in  Figures  39-45. 
Figure  39  shows  a  horizontal  bar  diagram.  It  will  be  noted 
that  no  scores  are  available  for  the  subject  on  a  number  of 
tests  listed  in  the  psychograph.  The  use  of  horizontal  bars 
which  are  not  joined  is  necessary  in  such  a  case,  since  a 
continuous  line  would  be  inapplicable.  The  boy  whose 
abilities  are  pictured  in  this  psychograph  exhibits  a  fairly 
clear-cut  tendency  to  excel  in  motor  performance.  His 
score  becomes  progressively  poorer  as  we  pass  from  the 
strictly  motor,  through  sensori-motor  and  perceptual,  to 
more  highly  "intellectual"  or  verbal  functions.  His  "men- 

1  Equal  units  can,  of  course,  be  obtained  by  the  use  of  scaling  techniques. 
This,  however,  is  a  laborious  procedure,  requiring  the  testing  of  a  large,  repre- 
sentative, and  normally  distributed  sampling  of  individuals.  The  question  of 
equal  units  is  no  more  essential,  furthermore,  for  the  measurement  of  trait 
variability  than  for  any  other  problem  involving  mental  tests.  It  has  been 
brought  up  in  this  connection  solely  because  of  a  rather  common  tendency 
erroneously  to  treat  transmuted  measures  as  equal  unit  scales. 




MENTAL  AGES.    (After  H.  L.  Hollingworth,  15,  p.  208.) 

tal  age"  on  the  different  tests  ranges  from  ir  years  in  a 
test  of  following  written  directions  to  slightly  over  18  in 
hand  grip. 

Figure  40  illustrates  the  use  of  a  circular  psychograph, 
plotted  in  terms  of  percentile  scores.  The  center  of  the 
circle  corresponds  to  a  zero  percentile,  the  outermost  cir- 
cumference to  100  percentile.,  and  the  middlemost  cor- 
responds to  an  average  performance.  The  radii  indicate 
the  individual's  relative  standing  in  different  functions. 
This  psychograph  shows  a  boy,  slightly  above  average 
in  intelligence,  who  is  fairly  uniform  in  most  of  the  traits 
measured.  With  one  exception,  his  percentile  ratings  vary 
between  45  and  70.  In  mechanical  aptitude,  this  boy  seems 
to  have  a  special  talent,  obtaining  a  percentile  score  of 
95  on  the  Stenquist  Mechanical  Aptitude  Test. 

Figures  41-45,  inclusive,  illustrate  the  continuous  line 
psychograph.  In  Figure  41  will  be  found  the  psychograph 





1.  Stanford -Binet 

2.  Trabue  Completion  Test 

3.  Cancellation 

4.  Digit-Symbol 

5.  Opposites 

6.  Stenquist  Mechanical  Ability  Test 

7.  Tonal  Memory 

8.  Pitch 

9.  Time 

10.  Intensity 

11.  Healy  Pictorial  Completion 

12.  Grip  in  Hand 

TERMS  OF  PERCENTILES,  (After  L.  S.  Hollingworth,  16,  p.  41.) 



Handwriting,  School  Marks__. 

Drawing,  School  Marks 

Drawing  Cart 

Drawing  Horse — 


Cancellation  Triangles 

Perceptual  Learning 

Lincoln  and  Pig,  Deferred 

Lincoln  and  Pig,  Immediate __ 

Marble  Statue P 

Aesthetic  Appreciation 

Code  1-g 

Punched  Holes 

Spatial  Relations 

Thurstone  Hand 

Thurstone  Reasoning 

Bonser  Reasoning 

Proverbs . 

Trabue  Completion . 

Word  Building- _ 

School  Standing. 

BinetLQ . . 



SPECIAL  TALENT  IN  REPRESENTATIVE  DRAWING.  (After  Manuel,  24,  p.  100.) 


of  a  girl  who  falls  within  the  lowest  10%  in  LQ.  and  in 
school  standing,  and  is  below  average  in  most  verbal  tests 
of  intellectual  functions.  But  she  exhibits  exceptional 
talent  in  representative  drawing.  When  tested,  this  girl 
was  in  the  sixth  grade  at  the  age  of  14.  Many  members  of 
her  family  were  reported  as  interested  and  talented  in 
drawing  or  painting.  Figure  42  shows  the  psychograph  of 


01Q|20[  3p[   40[  50 





Tonal  Memory 

Drawing  (Ratings) 

Grip  in  Hand 



General  Intelligence*. 
(Stanford-Binet;  Alpha) 



40    50 

70    80 


60    71 


90  100 

CIAL ABILITY  IN  REPRESENTATIVE  DRAWING.  (After  L.  S.  Holl  ing  worth,  16, 
P-  I7S-) 

a  14-year-old  girl  of  average  general  intelligence,  with 
exceptionally  high  ratings  in  music  and  drawing.  This 
subject  was  referred  for  psychological  examination  because 
she  was  doing  poorly  in  school.  She  had  been  attending  a 
superior  private  school  in  which  the  average  LQ.  was  120, 



a  fact  which  accounts  for  her  apparent  backwardness.  In 
this  school,  she  had  been  receiving  good  grades  only  in 
music  and  drawing.  In  Figure  43  is  given  the  psychograph 


010J  20    30    40[  50 

60|  70J  80|  90J100 

f  Pitch      -  -  













Tonal  Memory  



Drawing  (Ratings)  


Grip  in  Hand  




General  Intelligence- 
(Stanford-Binet;  Alpha) 


010|  20    30    40    50 

60J  70|  80  j  90J100 

16,  p.  179.) 

of  a  I  o-y ear-old  schoolboy.  On  intelligence  tests  this  boy 
ranked  close  to  the  looth  percentile,  having  an  I.Q.  of  151. 
In  reading,  arithmetic,  and  elementary  science,  his  school 
work  was  excellent.  In  music  tests,  on  the  other  hand,  he 
ranked  consistently  low;  his  music  teacher  regarded  him 
as  a  complete  failure  and  advocated  that  he  repeat  the 

Figures  44  and  45  have  been  plotted  in  terms  of  standard 
scores.   Both  are  taken  from  a  series  of  psychographs  con- 




Age  norm  for  6  years 







*         B 


SCHOOLGIRL.    (After  DeVoss,  8,  p.  350.) 

structed  by  DeVoss  (8)  on  children  who  had  been  selected 
for  their  high  intellectual  level.1  The  psychograph  in 
Figure  44  is  that  of  a  girl  with  a  Stanford-Binet  LQ.  of  192. 
This  child,  although  above  her  age  norm  in  all  of  the  tests 
but  one,  exhibits  marked  discrepancies  among  her  scores. 
She  is  highest  in  arithmetic  reasoning  and  also  shows  un- 
usual ability  in  tests  involving  reading  comprehension. 
In  four  information  tests  dealing  with  science,  language 
and  literature,  history  and  civics,  and  music  and  art,  she 
obtained  much  lower  scores.  In  the  last  named  test,  her 

1  Part  of  the  group  employed  by  Terman  in  his  extensive  investigation  OP 
gifted  children  (cf.  Ch.  XIII). 



score  was  even  slightly  below  her  age  norm.  In  Figure  45 
is  the  psychograph  of  a  schoolboy  with  an  LQ.  of  155,  who 
presents  a  very  different  mental  picture.  He  is  best  in 

SCHOOLBOY.    (After  DeVoss,  8,  p.  360.) 

music-art  information,  second  best  in  history-civics  in- 
formation, and  poorest  in  arithmetic  reasoning  and  com- 
putation. These  examples  illustrate  the  fact  that  intellec- 
tually superior  children,  although  above  their  age  norms  in 
most  mental  tests,  may  be  much  farther  above  average  in 
some  traits  than  in  others. 


The  term  A^tW  by  Hull 

(20)  to  designate  variability  from  ' 


individual.  This  is  to  be  distinguished  from  individual 
^^^^^.^hickjefers  to  the  differences  among  individuals 
in  a  single  trait.  All  of  the  statistical  techniques  commonly 
enapToyed  to  measure  the  amount  of  individual  variability 
can  be  applied  to  the  measurement  of  trait  variability, 
provided  that  the  scores  on  different  tests  are  expressed  in 
the  same  units. 

In  a  study  by  Hull  (20),  an  attempt  was  made  to  meas- 
ure the  amount  of  variability  in  different  traits  and  com- 
pare it  with  the  amount  of  individual  variability  on  single 
tests.  The  scores  of  107  high  school  freshmen  on  35  tests 
were  obtained.1  The  tests  included  several  subtests  from 
intelligence  scales,  as  well  as  tests  of  motor  characteristics, 
perception,  attention,  and  personality  traits.  Each  of 
the  35  sets  of  scores  was  transmuted  into  a  distribution 
with  an  average  of  81  and  an  S.D.  of  7  by  a  method  similar 
in  principle  to  that  employed  in  obtaining  standard  scores 
(cf.  Hull,  19).  The  particular  values  chosen  for  average 
and  S.D.  are  arbitrary  and  were  selected  chiefly  because 
of  their  similarity  to  school  grades. 

With  these  converted  scores,  the  S.D.  of  each  individ- 
ual's scores  in  the  35  tests  ivas  computed  as  a  measure  of 
thelimount  of  trait  variability.  There  were  thus  obtained 
as  many  S.D's  as  there  were  subjects,  viz.,  107.  T^^a^Gr- 

P'  P^£..T$wfo>  '4$  A$,s£  units 
employed,  represents  the  individual  variability.^  .any  QJie 

^"•UMMiMiiwllMttiJWMfci^  ,,«-,,;,.''-.*<,-  '  i      •     w  '     «r 

of^he  t^sts.  After  allowing  for  possible  chance  errors  of 
sampling  and  measurement, 

The..  exact  relationship  obtained  between  these  two  forms 
of  variability  is,  however,  affected  by  the  homogeneity  of 

1  The  original  data  were  collected  by  C.  E.  Limp  in  connection  with  an  in- 
vestigation of  shorthand  and  typewriting  aptitudes. 


the  group  employed.  The  more  homogeneous  the  group, 
the  smaller  will  be  the  original  S.D.'s  representing  in- 
dividual variability  in  raw  scores.  Because  of  the  trans- 
mutation technique  employed,  this  will  automatically 
make  the  deviations  within  the  individual  appear  larger. 
The  group  of  high  school  freshmen  upon  whom  Hull's 
data  were  obtained  was  a  fairly  representative  sampling, 
covering  a  wide  range.  His  results  are  therefore  typical  of 
what  would  be  found  on  many  groups.  A  universal  esti- 
mate of  the  amount  of  trait  variability  can  never  be 
arrived  at,  however,  because  such  a  measure  depends  in  a 
peculiarly  intimate  fashion  upon  the  nature  of  the  group 
within  which  it  is  obtained. 

ttl?  nevertheless  apparent  from  this  study  that  a  large 
amount  of  trait  variability  may  be  found  withia  a  group 
of  normal  individuals  who  were  not  selected  on  the  bans  of 
asymmetrical  development.  The  amount  of  trait  variability 
also  differed  with  the  individual.  Some  subjects  displayed 
much  more  uniformity  than  others  in  their  performance  on 
different  tests.  The  individual  S.D.'s  of  trait  variability 
ranged  from  4.3  to  9.09.  The  distribution  of  scores  of  each 
individual  on  the  35  tests  seemed  to  follow  the  general 
form  of  the  normal  curve.  MasJLof  Jthe  individual's  scores 
cj^stered  about  his  own  average,  only  a  few  scores  deviat- 
ing markedly  from  it  in  either  direction.  Fi#alit]2«*.fl&  re- 
of  ability  and  the  extent  of  This^tr|Lit^  variability.  There 

dividual  witt^^^ 

ual's  average  score  on  all  the  tests  f  and  Jii^trait^variability 

*ulM-^kl*-%*4''*'*'TJ  "*"""' 

coiiducted^  determine 

whether  gift^c^ldlW,,,M^,MlJ^BOTQ  specialized  in  their 

"^^^^^^^-^  •      p*u*«**.*^^^ 


intellect^lfcabniitie5«,tb.ai3...nGPmal  children.  A  group  of 
100  subjects  were  selected  on  the  basis  of  mental"  age  Jfrom 
a  larger  "group  of  "gifted  children"  employed  by  Terman 
(cf.  Ch.  XIII).  The  mental  ages  of  DeVoss'  group  ranged 
between  14  and  15-5,  with  an  average  of  14-8;  the  chrono- 
logical ages  ranged  from  8-6  to  n-i.  The  average  LQ. 
was  149.4,  and  the  range  from  136  to  1  80.  In  school  grade, 
the  children  were  scattered  from  the  third  to  the  eighth 
grade,  inclusive.  Tb^se^^  ,with-ra 

^of^  unselected  eighth  grade  schoolchildren 

ages  as,  the,  superior 
group.  Both  groups  were  given  the  Stanford  Achievement 
Test,  consisting  of  seven  subtests  on  different  school  sub- 
jects, as  well  as  information  tests  in  special  fields.  All 
scores  were  reduced  to  standard  measures. 

within  each  sub- 
.differences  .large  enough  to 

By  means  of  a  specially  devised  statistical 
formula,1  it  was  possible  to  estimate  how  large  a  trait 
difference  might  be  obtained  simply  through  the  opera- 
tion of  chance  factors,  such  as  inadequacies  in  the  tests 
employed.  Upon  the  application  of  this  formula,  it  was 
discovered  that  a  large  percentage  of.  the  trait  . 
felTbe^nd  the  chanq^Jiiaits,  and.mpst  t 
a  triiejiisxxrepancy  in  the  individual's  staading,in4he  traits 
compared.  In  Table  XIII  are  given  the  percentage  of  trait 
differences,  in  both  the  gifted  and  control  groups,  which 
fell  outside  of  the  distribution  of  differences  expected  by 
chance.  The  percentages  in  the  gifted  group  are  given 
above  the  diagonal,  those  in  the  control  group  below  the 
diagonal  The  tests  which  are  being  compared  are  in- 
dicated in  the  top  row  and  first  column  to  the  left.  Thus, 

1  A  formula  for  the  computation  of  the  P.E.  of  the  difference  between  an 
individual's  standard  scores  on  any  two  tests. 



in  the  gifted  group,  24%  of  the  differences  between  scores 
on  arithmetic  reasoning  and  computation  fell  beyond  the 
chance  limits ;  the  corresponding  percentage  for  the  control 
group  is  34%,  and  so  on. 





(After  DeVoss,  8,  p.  325) 

Tests  to  Be 


.&  ^ 

1  *s 
•1  5 




§  s 

tf  ? 


s   v> 

txo   txo 



*-    o 





























































v  33 



































*  Gifted  group  above  the  diagonal:  upper  right-hand  block.   Control  group 
below  the  diagonal:  lower  left-hand  block. 

It  will  be  noted  that  in  every  test  pair  compared  there 
are  found  differences  over  and  above  those  expected  by 
chance.  This  is  true  of  both  gifted  and  control  groups. 
The  percentages  of  such  differences  are  also  closely  similar 
in  the  two  groups,  test  by  test.  In  the  gifted  group,  the 


percentages  vary  from  13  to  37,  in  the  control  group  from 
13  to  40.  The  average  percentages  are  28.89  an<3  27.82 
for  gifted  and  control  groups,  respectively.  Out  of  the 
28  inter-test  comparisons  given  in  Table  XIII,  the  gifted 
group  has  the  larger  percentage  of  excess  differences  13 
times,  the  control  group  has  the  larger  percentage  12 
times,  and  the  two  have  identical  percentages  in  3  cases. 
Thus  t  there  .seems  to  be  no  appreciable  or  consistent  dif- 
ference in,  trait  variability  between  intellectually  normal 
and  superior  children. 


The  examination  of  extreme  examples  of  asymmetrical 
t,  _  as,  well  as  the  measurement,  of  .trait  variabil- 
in  general,  suggests  that.  superior 
may  be  associated  with  inferior  abilities 

injDther  respects.  It  is  not  to  be  concluded  from  this, 
^SSZSIiJliS^  ^PmPensa^on  is.  the  rule.  Superior  standing 
in  one  trait  does  not  imply  inferiority  in  another.  We  have 
cited  only  examples  in  which  individuals  with  a  high 
standing  in  a  certain  trait  A  make  a  poor  showing  in  a 
second  trait  B.  We  could  with  equal  facility  find  cases  in 
which  the  individual  is  superior  in  A  as  well  as  J?,  or  supe- 
rior in  A  and  average  in  B.  TJbis^juxi  act3  is  wjiat  ,w,^n 
by^jL????  SSn^i^on-  If  *7  arious  a^llti&s  jM:e..spedfi<s«a.nd 
mutually  independent,  so  that  an  i^dividuallso&taadii^  in 

,„,„>,.  „,«  Mvu  ^^^^ml^m^m^^a^''^  ,  TOl-V<v*"rP"t^*^*.«  «•»-!*""»""«'   ""'  -"'  '  ° 

onejtells  usjaothing  about  his  relative  standing,m  aaajher, 
we  should  expect  the  correlation  between  such  jtbiljties  to 
be  zero*  "or"  very  low."  Correraflon^tRus  offers  another  ap- 
proac"h"to  the  analysis  of  trait  variability.  We  may  now 
turn  to  a  consideration  of  some  typical  correlational  results 
on  the  various  traits  suggested  by  case  studies  and  psy- 
chographic  analysis. 


findings  with  the  Hevner  tests  of  music  appreciation  are 
illuminating.  In  a  group  of  74  college  students,  Hevner 
(13)  obtained  a  correlation  of  -.17  between  one  form  of 
her  music  appreciation  test  and  the  Minnesota  College 
Ability  Test.  The  latter  is  very  similar  in  its  content  to 
current  intelligence  examinations  adapted  to  the  college 
level.  In  the  music  appreciation  test,  four  versions  of 
parts  of  24  musical  compositions  are  rendered  on  the  piano. 
These  versions  include  the  original  and  four  distortions. 
One  is  a  simplification  of  the  music  in  which  many  of  the 
tones  are  omitted,  so  that  the  music  sounds  bare,  hollow, 
and  uninteresting.  In  a  second  version,  the  music  is 
elaborated  by  introducing  additional  tones  into  the  chords 
and  by  ornamentation,  so  that  the  piece  is  overdone,  con- 
fusing, and  "frilly."  In  a  third  version,  the  melodic  pro- 
gression or  phrasing  is  altered,  the  music  being  thereby 
rendered  formless  and  unbalanced.  The  subject  is  re- 
quired to  state,  in  each  case,  which  of  the  four  versions  he 
prefers.  This  test  has  been  found  to  be  quite  diagnostic  of 
musical  aptitude,  correlating  highly  with  other  music  tests 
and  independent  criteria. 

The  same  subjects  were  given  another  form  of  the  music 
appreciation  test  which  included  48  items  presented  in 
only  two  versions,  the  original  and  one  distortion.  The 
correlation  between  intelligence  test  performance  and 
scores  on  this  test  proved  to  be  —.15.  The  low  negative 
correlations  of  these  tests  with  intelligence  would  suggest 
a  slight  tendency  for  the  more  "intelligent"  subjects  to  be 
inferior  in  music  appreciation.  The  correlations  are  too 
low,  however,  to  indicate  a  significant  trend  and  the  two 
traits  can  be  regarded  as  practically  unrelated. 

Intelligence  has  also  been  found  to  have  little  or  no 
relationship  with  performance  on  the  Seashore  tests  of 
musical  sensitivity.  In  an  investigation  (9)  on  230  college 


students,  LQ.'s  were  estimated  from  scores  on  the  Otis 
Self-Administering  and  Army  Alpha.  These  I.Q.'s  were 
correlated  with  scores  on  each  of  the  Seashore  tests,  with 
the  following  results: 

Musical  Test  Correlation  with  LQ. 

Pitch  .32 

Intensity  .01 

Time  .13 

Consonance  .09 

Tonal  memory  .10 

Rhythm  .12 

Aptitude  in  pictorial  art  shows  a  similar  independence 

'•^L^^^^ ^^^^jwaww— mi*M&*~M*HH,uMi>***»f'«'»>ti*f  t*-«*i"<    '  "  ~"   '"•    " '"   "    "       "•"'•""'    '"*>•"""'      "  "      <  <J/--"-         .        .  iw.,¥1     ,    „,,, 

of  general  jntdligence.,  Meier  (25)  obtained  a  correlation 
oCz^ytfi Jpetween  the  Terman  Group  Test  of  Intelligence 
and  an  art  judgment  test,  within  a  group  of  55  high  school 
sSHentsl'  Tfie  same  test  correlated  —.018  with  Thorndike 
Intelligence  Test  among  53  college  students.  Both  correla- 
tions are  sufficiently  low  to  indicate  an  absence  of  relation- 
ship between  the  two  traits.  The  art  judgment  test  con- 
sisted of  a  series  of  pairs  (or  larger  groups)  of  pictures 
reproduced  in  black  and  white.  Within  each  pair  (or 
group),  one  picture  was  the  original,  the  other  (or  others) 
a  variation  which  disrupted  the  organization  of  the  picture 
by  altering  its  symmetry,  balance,  harmony  and  unity, 
rhythm,  or  similar  features.  The  subject  was  to  indicate 
which  picture  he  preferred  in  each  case. 

In  the  effort  to  discover  the  relationship  between  draw- 
ing ability  and  other  traits,  Ayer  (3)  conducted  an  in- 
vestigation on  51  high  school  students.  The  subjects  were 
shown  a  turkey  feather  and  were  asked  to:  (i)  make  a 
representative  drawing  of  it,  (2)  diagram  it,  and  (3)  de- 
scribe it  verbally.  Twenty-four  hours  later,  they  were 
again  required  to  draw  a  diagram  of  the  feather  from 
memory  and  also  to  answer  certain  questions  about  it 


(retention  test).  Each  subject's  performance  on  all  five 
tests  was  scored  independently  by  10  judges  in  order  to 
arrive  at  a  final  estimate  of  his  score.  The  various  correla- 
tions computed  among  these  scores  are  reported  below. 

Representative  drawing  and  verbal  description  —.271 

Representative  drawing  and  diagramming  —.052 

Diagramming  and  verbal  description  .231 

Representative  drawing  and  retention  —.022 

Verbal  description  and  retention  .234 

Diagramming  from  memory  and  retention  .433 

As  would  be  expected,  the  only  positive  correlations  are 
those  between  diagramming  (immediate  or  delayed), 
verbal  description,  and  retention.  These  tasks  have  a 
good  deal  in  common,  since  they  all  involve  careful  ana- 
lytic observation  of  the  object.  Little  or  no  artistic  talent 
is  required  in  diagramming.  The  only  task  which  implies 
drawing  ability  to  any  appreciable  extent  is  representative 
drawing.  This,  it  will  be  noted,  shows  very  little  correla- 
tion with  the  other  tasks.  All  the  correlations  computed 
by  Ayer  with  representative  drawing  are  negative.  Two 
are  too  low  to  indicate  the  presence  of  any  relationship; 
the  third,  although  also  low,  suggests  the  possibility  that 
those  subjects  who  rendered  a  good  representative  draw- 
ing may  have  overlooked  certain  details  and  concentrated 
chiefly  on  a  general  impressionistic  observation. 

Mechanical  fa^litod^&^  spmd^b  jlity. 

Stenquist  reports  a  correlation  of  .330  between  the  Sten- 
quist  Assembling  Test  .and ..a, .,.,ccanp,Q$ite. t score, .froin^six 
verbal  intelligence  tests  in  a  .group  of  267,,  seventh  and 
eighth  grade  boys.  Although  reliably  higher  than  zero, 
this  correlation  indicates  only  a  very  slight  degree  of  re- 
lationship. The  Stenquist  Assembling  Tests  involve  the 
construction  of  common  mechanical  objects  such  as  a  lock, 
bicycle  bell,  chain,  and  mouse  trap,  from  given  parts. 


In  a  very  extensive  and  thorough  investigation  on 
mechanical  aptitude  conducted  at  the  University  of 
Minnesota  (cf.  29)5  various  correlations  were  computed 
between  mechanical  tests  and  intelligence.  I-Q-'s  were 
estimated  from  scores  on  the  Otis  Intelligence  Test.  ..The 
correlation  between  these  I.Q/s  and  a  mechanical  aptitude 
battery,1  composed  exclusively  of  apparatus  or  manipula- 
tion tests,  proved  to  be  only  .13  within  a  group  of  100 
Junior  High  SchooTbo^^  ^  obtaineci  a  correla- 

tion of  .07  between  a  vocabulary  test  and  the  Minnesota 
Paper  Form  Board  Test  among  225  male  college  students. 
The  latter  is  a  paper-and-pencil  test  of  the  ability  to 
handle  spatial  relations.  Vocabulary  tests,  which  measure 
the  subject's  understanding  of  word  meanings,  have  been 
found  to  correlate  so  highly  with  the  majority  of  common 
intelligence  tests  as  to  be  practically  interchangeable 
with  them.  From  these  examples,  it  is  apparent  that  in 
large  groups  of  subjects  of  different  age  and  academic 
level,  only  a  very  low  positive  correlation  exists  between 
spatial  or  mechanical  ability  and  the  verbal  type  of  in- 
telligence test.  When  the  mechanical  problems  are  pre- 
sented verbally,  the  correlations  with  intelligence  tests 
are  usually  higher  because  of  the  common  influence  of  the 
comprehension  of  verbal  directions,  knowledge  of  words, 
and  general  facility  with  verbal  material. 

Of  the  special  aptitudes  suggested  by  case  studies,  only 
numerical  ability  remains  to  be  examined.  In  spite  of 
the  indisputable  presence  of  "mathematical  prodigies" 
who  are  deficient  in  other  respects,  numerical  ability  has 
not  usually  be^n^k^^^^ 

metic  tests  are  also  frequently  inciudedj^iot^ligence 
scalSsnCecent  correlational  analysis  has  demonstrated, 

1  A  series  of  tests  each  of  which  is  given  an  optimum  weight  and  combined  so 
as  to  give  the  best  possible  estimate  of  the  trait  measured. 


however,  that  the  relationships  between  verbal  and  nu~ 
merical  "tests*  "are  much'  lower  than  those  within  either 
group  "of  Abilities.  In  many  investigations,  the  correla- 
tions Between  verbal  and  numerical  tests  were  no  higher 
than  those  between  verbal  tests  and  the  various  special 
aptitudes  discussed  above. 

obtained  a  correlation  of  .09  between  a  read- 
n  test  and  a  test  of  arithmetic  reasoning 

in  jajjroup  of  140  seventh  grade  children.  Schneck  (30) 
tested  210  college  men  with  5  verbal  and  4  numerical  tests. 
The  average  of  the  correlations  between  all  possible  pairs 
of  verbal  tests  was  4920;  the  corresponding  average 
correlation  for  the  numerical  tests  was  .3383.  When  ver- 
bal and  numerical  tests  were  paired  off,  the  average  of 
the  correlations  so  obtained  was  only  .1441.  Even  this 
rather  low  positive  correlation  probably  resulted  from 
the  fact  that  in  at  least  one  of  the  numerical  tests,  the 
problems  were  expressed  in  verbal  terms,  and  in  all  of  the 
tests  the  directions  were  given  verbally.  Anagtasij^)  found 
a  correlation  of  —.or  between  arithmetic  reasoning  and 
vocabulary  among  225  male  college  students.  In  a  sub- 
sequent investigation  by  the  same  author  (2)  on  140  col- 
lege women,  all  possible  correlations  were  computed 
among  two  verbal  and  two  numerical  tests.  The  correla- 
tion between  the  two  verbal  tests  (vocabulary  and  analo- 
gies) was  .65,  and  that  between  the  two  numerical  tests 
(arithmetic  reasoning  and  number  series  completion)  .58. 
The  average  of  four  correlations  obtained  by  pairing  each 
of  the  verbal  tests  with  each  of  the  numerical  was  only 
.16.  These  various  findings  suggest  that  verbal  and 

^WiMMI»IMK»*lm«NWWMK.M»««^^  *  ,  fflMMSWlWl**  «#"!*  "''*"  '  '*"*   *  """'  *™  "'*"  '  **^*J"**'"1WW»*'*"11  "'"«»  •**»•«     >  *"*'  >»**  '•  *"<"  '  **" 

to  be  n^suri^,^ft.J*8p^ciaLapti- 

sense^jsas_,the  Bother,  tests,  jdi^cussed 

No  mention  of  personality  traits  has  been  made  in  this 


chapter.  This  is  because  the  independence  of  personality 
and  intellectual  traits  is  now  quite  generally  recognized. 
Emotional  instability,  for  example,  may  be  found  over  a 
wide  range  of  intellectual  levels.  Insanity  among  highly 
gifted  persons  is  not  unknown.  Extreme  emotional  dis- 
orders are  perhaps  not  as  common  among  the  intellectu- 
ally ablest  as  among  more  mediocre  individuals  because 
the  former  can  exert  somewhat  more  control  over  social, 
economic,  and  other  aspects  of  their  environment  and  are 
therefore  less  frequently  subjected  to  conflicts  or  frustra- 
tions. At  the  same  time,  certain  forms  of  insanity  are 
rare  among  low  grade  mental  defectives  because  these 
individuals  have  too  restricted  a  mental  life  to  experience 
much  stress  or  strain. 

^2£!^£J^^  a^so  to  ^e  y^T 

largely  independent  of  ability.      Criminals  are  not  sig- 

nificantly differentiated  from  non-criminals  in  mtellegtual 
statusT  ThTs  waV  demonstrated  in  an  extensive  investiga- 
tion by  Mucdbis^  Alpha  was 

admin^  in   the 

penitentiaries  of  ^  five  states.  The  distribution  of  fc  the 
gco^^TtEiJ^rou£  coincided  ye^  closely  with  that^of  the 
ar^^drpt  from  the  same  five  states,  j^mong^  juvenile 

delinquents,   ^ajy^JS^^  *n~ 

tellectuaf  (deviation  when  thejsubjects  are  .j:jompared  with 

a  -iio53e]^ 

level^ET^  In  the  course  of  extensive 

experimentation    with    large    groups    of    schoolchildren, 

velT  122L£22£lS^ons 

traits.     The  chifdren  were  tested  in  practical  everyday 
situations   and   without  their  knowledge.      Among  the 

1  Separate  investigations  were  also  conducted  on  smaller  groups  of  foreign- 
bora,  Negro,  and  women  prisoners. 


character  traits  investigated  were  lying,  cheating,  steal- 
ing, charity,  cooperativeness,  persistence,  and  inhibition. 
When  the  questionnaire  type  of  test  was  employed  (cf.  12), 
the  correlations  with  intelligence  were  much  higher,  owing 
to  the  common  dependence  of  such  tests  upon  verbal 

The  administration  of  tests  of  social  traits  or  emotional 
instability  to  large  groups  also  reveals  little  or  no  rela- 
tionship with  intelligence-  Thus,  for  example,  no  appre- 
ciable correlation  has  been  found  between  intelligence 
test  score  and  performance  on  the  Colgate  Tests  of  Emo- 
tional Outlets  for  measuring  introversion  and  emotional 
instability.  Within  a  group  of  218  college  students,  a 
correlation  of  .02  was  obtained  between  introversion  scores 
and  performance  on  the  Thorndike  Intelligence  Test. 
The  correlation  between  scores  on  the  same  intelligence 
test  and  number  of  neurotic  symptoms  proved  to  be  .008 
among  203  male  subjects  and  —.12  among  women  (num- 
ber not  stated).  Thurstone  (34)  reports  a  correlation  of 
.037  between  scores  on  his  Neurotic  Inventory  and  the 
American  Council  of  Education  intelligence  test  among 
694  college  freshmen.  Bender  (5)  found  a  correlation  of 
.0008  between  scores  on  the  same  intelligence  test  and 
performance  on  the  Allport  Ascendance-Submission  test 
in  a  group  of  192  college  sophomores.  It  is  almost  super- 
fluous to  discuss  correlations  between  intelligence  and 
personality  tests^  since  in  the  construction  of  the  latter  a 
definite  attempt  is  made  to  avoid  an  appreciable  correlation 
'with  intelligence.  The  assumption  is  implicit  in  techniques 
of  personality  test  construction  that  the  traits  to  be 
measured  are  independent  of  intellectual  status.  That 
these  efforts  have  usually  succeeded  testifies  to  the  sound- 
•  ness  of  the  assumption. 



1.  Anastasi,  A.     "A  Group  Factor  in  Immediate  Memory,'5 
Arch.  Psychol,  No.  120,  1930.   Pp.  61. 

2.  .     "Further  Studies  on  the  Memory  Factor,"  Arch. 

Psychol,  No.  142,  1932.  Pp.  60. 

3.  Ayer,  F.  C.    The  Psychology  of  Drawing.    Baltimore:  War- 
wick and  York,  1916.  Pp.  186. 

4.  Barr,  M.  W,    "Some  Notes  on  Echolalia,"  /.  Nerv.  Mental 
Diseases,  1898,  25,  20-30. 

5.  Bender,  I.  E.    "Ascendance-Submission  in  Relation  to  Cer- 
tain Other  Factors  in  Personality,"  /.  Abn.  Psychol,  1928— 
29,  23,  I37-I43- 

6.  Binet,  A.      Psychologie  des  grands  calculateurs  et  joueurs 
d'echecs.  Paris:  Hachette,  1894.  Pp.364. 

7.  Bronner,  A.  F.    The  Psychology  of  Special  Abilities  and  Dis- 
abilities. Boston:  Little,  Brown,  1919.  Pp.  269. 

8.  DeVoss,  J.  C.     "Specialization  of  the  Abilities  of  Gifted 
Children,"  in  Genetic  Studies  of  Genius,  Terman,  L.  M.,  ed. 
(Vol.  I,  Ch.  XII).    Stanford  Univ.,  Calif.:  Stanford  Univ. 
Press,  1925.  Pp.  648. 

9.  Fracker,  C.  C.,  and  Howard,  V.  M.    "  Correlation  between 
Intelligence  and  Musical  Talent  among  University  Stu- 
dents," Psychol  Mon.,  1928,  39,  157-161. 

to.  Hartshorne,  H.,  and  May,  M.  A.   Studies  in  Deceit.   N.  Y.: 

Macmillan,  1928.   Book  I,  pp.  414;  Book  II,  pp.  306. 
[I.  Hartshorne,  H.,  May,  M.  A.,  and  Mailer,  J.  B.    Studies  in 

Service  and  Self-Control   N.  Y.:  Macmillan,  1929.    Pp.  559. 
[2.  Hartshorne,  H.,  May,  M.  A,  and  Shuttleworth,  F.  K.  Studies 

in  the  Organization  of  Character.   N.  Y.:  Macmillan,  1930. 

Pp.  503. 
[3.  Hevner,  K.     "A  Study  of  Tests  for  the  Appreciation  of 

Music,"  /.  Appl  Psychol,  1931,  15,  S7$~5%3* 

14.  Hoitsma,  R.  K.,  "Reliability  and  Relationships  of  the  Col- 
gate Mental  Hygiene  Tests,"  /.  Appl  Psychol,  1925,  9, 


15.  Hollingworth,  H.  L.     Judging  Human  Character.     N.  Y.: 
Appleton,  1922.  Pp.  268. 


16.  Hollingworth,  L.  S.     Special  Talents  and  Defects.     N.  Y.: 
Macmillan,  1925.  Pp.  216. 

17.  .    Gifted  Children:  Their  Nature  and  Nurture.    N.  Y. : 

Macmillan,  1926.  Pp.  374. 

18.  .  " The  Musical  Sensitivity  of  Children  Who  Test  above 

135  I.Q.,"  J.  Educ.  PsychoL,  1926,  17,  95-109. 

19.  Hull,  C.  L.     "The  Conversion  of  Test  Scores  into  Series 
Which  Shall  Have  Assigned  Mean   and   Degree  of  Dis- 
persion," /.  Appl.  Psychol.,  1922,  6,  298-300. 

20.  .  "Variability  in  Amount  of  Different  Traits  Possessed 

by  the  Individual,''  /.  Educ.  PsychoL,  1927,  18,  97-104. 

21.  Kelley,  T.  L.    Statistical  Method.    N.  Y.:  Macmillan,  1924. 
Pp.  390- 

22.  .     Crossroads  in  the  Mind  of  Man.     Stanford  Univ., 

Calif.:  Stanford  Univ.  Press,  1928.  Pp.  238, 

23.  Lindley,  E.  H.,  and  Bryan,  A.  L.    "An  Arithmetical  Prod- 
igy," Psychol.  Rev.,  1900,  7, 135. 

24.  Manuel,  H.  T.  A  Study  of  Talent  in  Drawing.  Bloomington, 
III:  Public  School  Pub.  Co.,  1919.   Pp.  152. 

25.  Meier   N.  C.     "^Esthetic  Judgment  as  a  Measure  of  Art 
Talent,"  Univ.  Iowa  Stud. :  Series  on  Aims  and  Progress  of 
Research,  1926,  I,  No.  19.  Pp.  30. 

26.  Minogue,  B.  M.    "A  Case  of  Secondary  Mental  Deficiency 
with  Musical  Talent,"  /.  AppL   PsychoL,   1923,   7,  349- 

27.  Mitchell,   F.   D.      "Mathematical   Prodigies,"   Amer.   J. 
Psychol.)  1907,  18,  61-143. 

28.  Murchison,  C.     Criminal  Intelligence.     Worcester,  Mass.: 
Clark  Univ.  Press,  1926.  Pp.  291. 

29.  Paterson,  D.  G.,  et  al.    Minnesota  Mechanical  Ability  Tests. 
Minneapolis:  Univ.  Minn.  Press,  1930.   Pp.  586. 

30.  Schneck,  M.  M.  R.     "The  Measurement  of  Verbal  and 
Numerical  Abilities,"  Arch.  PsychoL,  No.  107,  1929.  Pp.  49. 

31.  Scripture,  E.  W.  "Arithmetic  Prodigies,"  Amer.  J.  PsychoL, 
1891,  4,  1-59. 

32.  Slawson,  J.     The  Delinquent  Boy.     Boston:  Badger,  1926. 
Pp.  477. 


33.  Stenquist,   J.    L.      Measurements   of  Mechanical  Ability. 
N..Y.:  Teachers  College,  Columbia  Univ.,  1923.  Pp.  101. 

34.  Thurstone,  L.  L.,  and  T.  G.     "A  Neurotic  Inventory," 
/.  Soc.  PsychoL,  1930,  i,  3-30. 

35.  Tredgold,  A.  F.    Mental  Deficiency.    N.  Y.:  Wood,  1922. 
Pp.  569. 


The  lack  of  relationship  between  intelligence  test  per- 
formance and  several  mental  functions  (cf.  Ch.  X)  raises 
a  question  as  to  what  constitutes  "intelligence."  It  will 
be  recalled  that  the  original  purpose  of  intelligence  tests 
was  to  sample  a  large  number  of  different  abilities  in 
order  to  arrive  at  an  estimate  of  the  subject's  general 
level  of  performance.  Insofar  as  A^mdiyjduaJ'.s,  sjtanding 
in  specific  functions  differs,  , such,, a  general  , estimate  is 
\nsatisfactory.  It  is  apparent,  however,  that  current  in- 
telligence tests  do  not  even  furnish  an  adequate  estimate 
of  the  average  ability  of  the  individual,  since  they  are 
overweighted  with  certain  functions  and  omit  others. 
Thus  in  the  non-language  and  performance  tests  of  in- 
telligence, spatial  aptitude  plays  the  dominant  role. 
Most  paper-and-pencil  tests,  on  the  other  hand,  measure 
chiefly  verbal  ability  and,  to  a  slighter  extent,  numerical 

Sincej:heJ^  of  test  is  by  far  the ,,most  frequently 

emgjgxed^^the  term  "intelligQace'!,  has  come  to" be  used 
alm^stm£ynonymously  with  verbal  ability.  Mental  age 
on  the  Stanford-Binet,  for  example,  has  been  found" to 
correHFe"^^^^  performance  on  the  vo- 

caBHagr  test  of  the  scale  (cf.  28).  This  finding  led  Terman 
to  suggest  that  "a  mental  age  based  on  the  vocabulary 
score  alone  would  not  be  far  wrong  in  a  large  per  cent  of 
the  cases"  (30,  p.  454). 

From  another  angle,  most  jntellig^c^^^estsftmaY  be 
regarded  as  measures  of  scholastic  aptitude,  or  abtiity^to 

"•"•->•"»  .-w,,  i'       J         <         '  '  '      *'"""*"•"   "''^wi-  «».%,  np^^^M*,    **.»,         * 

*""""""'"—*  -  -"***    *'   "  "  296 


succeedjn^our  schools.  This  is  illustrated  particularly  well 
by  the  procedure  commonly  followed  in  validating  in- 
telligence tests.  The  term  validity  is  used  to  denote  the 
degree  to  which  a  test  actually  measures  what  it  purports 
to  measure.  In  the  case  of  intelligence  tests,  validity  is 
usually  checked  against  school  success  as  a  criterion. 
Scores  on  the  test  are  correlated  with  school  grades  or 
teachers'  estimates  of  ability,  and  the  higher  these  correla- 
tions the  more  valid  the  test  is  said  to  be. 

Some  rather  illuminating  data  on  the  relationship  be- 
tween intelligence  test  performance  and  scholastic  ability 
were  brought  together  bjULdPkj*-  (17,  Ch.  VIII).  Avail- 
ing himself  of  material  gathered  by  several  investigators, 
Kelley  presents  correlations  between  scores  on  several 
intelligence  tests  and  performance  on  the  Stanford 
Achievement  Test.  The  latter  is  a  carefully  constructed 
and  standardized  test  covering  several  school  subjects 
and  was  designed  for  use  in  place  of  ordinary  school 
examinations.  The  correlation  of  this  examination  with 
the  National  Intelligence  Test  proved  to  be  .66;  with 
Otis  Intermediate  Test  of  Intelligence,  the  correlation 
was  .79;  and  with  Illinois  General  Intelligence  Test,  it  was 
.71.  These  correlations  are  practically  as  high  as  those 
obtained  when  different  intelligence  tests  are  correlated 
with  each  other.  Thus  Stanford-Bract  and  National  In- 
telligence Test  correlated  .84,  National  Intelligence  Test 
and  Terman  Group  Test  .79,  and  Stanford-Binet  and 
Terman  Group  Test  .75. 

^  it  would  seem  that  "in- 

misleading,  term 

.  Recently,  there  has  been  a  tendency  eitKeFlo 
discard  or  to  qualify  the  blanket  term  "intelligence"  and 
to  give  more  specific  and  more  informative  names  to 
mental  tests.  Thus  we  find  the  Intelligence  Scale  CAVD, 


named  after  its  four  component  parts.  Completion, 
Arithmetic,.  Vocabulary,  and  Directions;  this  test  claims 
to  measure  no  more  than  the  "CAVD  segment"  of  in- 
tellect (cf.  41).  The  Minnesota  College  Ability  Test  and 
the  Scholastic  Aptitude  Test  constructed  by  the  College 
Entrance  Examination  Board  are  also  examples  of  the 
explicit  recognition  of  the  specific  nature  of  the  measuring 
instrument.  In  the  latter  test,  the  subject  is  now  given  a 
separate  score  in  the  verbal  and  numerical  parts  (cf.  21), 
a  procedure  which  testifies  further  to  the  influence  of 
studies  of  trait  relationship  upon  test  construction  and 
interpretation.  Thus  emphasis  is  shifting  from  indices  of 
so-called  general  intelligence  to  concrete  behavior  traits 
and  their  interrelations. 

Such  interrelationships  among  an  individual's  abilities 
are  included  under  the  rubric  of  mental  organization.  In 
investigating  this  problem,  we  are  not  only  concerned 
with  the  establishment  of  the  presence  or  absence  of  trait 
variations,  but  we  are  also  interested  in  an  exact  formula- 
tion of  the  principles  underlying  the  amount  and  di- 
rection of  such  variations.  Through  such  an  analysis, 
we  shall  be  able  to  identify  the  basic  components  or  uni- 
tary traits  into  which  the  individual's  behavior  may  be 

Theories  of  mental  organization  are  very  old.  As  long 
as  philosophers  have  discussed  the  nature  of  mind,  they 
have  proposed  theories  to  explain  how  the  "parts"  of  the 
mind  were  related  or  organized.  With  these  speculations, 
however,  we  are  not  concerned.  It  is  only  since  the  appli- 
cation of  mental  tests  and  quantitative  methods  that  the 
relationships  among  the  varied  responses  of  the  individual 
could  be  measured.  The  more  recent  theories  have  been 
offered  as  interpretations  of  specific  evidence  and  hence 
have  a  more  objective  foundation. 



The  problem  of  mental  organization  was  first  placed 
upon  an  "empirical  basis  with  the  publication  of  Spear- 
man's 1904  article  (26)  in  which  were  presented  a  theory 
and  a. new  method  of  investigation.1  According  to  Spear- 
man^^Tzvo-Factor  Theory,  2  all  intellectual,  activities 
have  Jnfc  common  one  fundamental  function  which,  is 
caliei,tiie~~G^<?f al  factor.,  or  g.  In  addition,  each  activity 
has.  Specific^®?  s  factors.  The  s  factors  are  exceedingly 
numerous  and  strictly  specific  to  each  activity  of  the 
individual.  No  two  activities  can  share  specific  factors, 
by  definition.  Spearman  argued  that  such  a  theory  Js 
cpusktent  withfico^^  Thus  the  presence 

of  different  specifics  in  every'  activity  would  explain  the 
absence  of  perfect  +  i.oo  correlations;  no  two  activities, 
however  much  they  may  depend  upon  the  g  factor,  are  en- 
tirely free  from  specifics.  The  fact  that  most  abilities  are 
positively  correlated,  on  the  other  hand,  is  attributed  to 
the  ubiquitous  g.  Different  proportions  of  g  and  s  in  each 
activity  wpulcl  prpduce  a  wide  range  of,  positive  correla- 
tions,., all  higher^  than  zero  and  lower  than  i.oo. 

Itjollows  from  the  Two-Factor  theory  that  the  aim  of 
mental  ^tegtmgLiliouldL  |>£;tQ  m^as,nr^  the  amount  of  each 
iBdisddjualls  g.  If  this  factor  runs  through  all  abilities,  it 
furnishes  the  only  basis  of  prediction  of  the  subject's 
performance  from  one  situation  to  another.  It  would  be 
futile  to  measure  specific  factors,  since  each  operates  in 
only  one  activity.  Accordingly,  Spearman  proposed  that 
a  single, tesvJugbJy'." 

1  Discussed  below.    • 

2  For  a  discussion  of  the  main  points  of  this  theory  and  its  modifications,  see 
27  and  28. 

3  Spearman  employs  the  term  "saturation"  to  denote  the  degree  to  which  an 
activity  depends  upon  the  g  factor. 


for  the  heterogeneous  collection  of  items  in  intelligence 
scales.  He  suggested  that  tests  dealing  with  abstract  rela- 
tions, such. .-is;  the  analogies  test,  are  the  best  single 
measures  o£*&  and  could  therefore  be  employed  for  this 
purpose.  '-  --  ^,i 

In  regard  to  the  nature- of  g,  Spearman  offers  only  ^tenta- 
tive suggestions/^  He^prpposes  that  g  may  be  regarded  j,s 
the  genei^J  ^'mental  'ef^rgy  of  the  individual  and  the  s 
factors  asTtiie  "engines"  through  which  it  operates,  or 
the  specific^  neurone  Bitterns  involved  in  each  activity. 
This  interpretation  of  |  and  s  is  not,  however,  an  integral 
nor  a  basic  part  of  the  Two-Factor  theory.  It  might  be 
noted  that  -Spearman's  g  would  also  furnish  a  basis  for 
the  popul^  jnqtiqn  of  general  intelligence. 

Even  from  the  outset,  Spearman  realized  that  the  Two- 
Factor  theory  must  be  qualified.  When  the  activities 
compared  are  very  similar,  a  certain  degree  of  correla- 
tion may  result  over  and  above  that  attributable  to  the 
g  factor.  Thus  in  addition  to  the  general  and  specific 
factors,  there  might  be  another  intermediate  class  of 
factors,  not  so  universal  as  g  nor  so,  strictly  specific  as  the 
j,, factors.  Sj£}ijs, facto^ ..which 4s.  common,  tp  a  group  of 
activities  but,  not.  to  all,  has  been 'designated  a, .group 
factor.  In  the  early  formulation  of  his  theory,  Spearman 
admitted  the  possibility  of  very  narrow  and  negligibly 
small  group  factors,  following  subsequent  investigations 
by.sey.ejal  of  his  students,  he  included  much  broader,  group 
factors  of  verbal  ability  and  spatial  or  "practical"  ability. 

Finally,  on  the  basis  of  a  series,, of  studies,,  additional 
general  factQ^s  )y,gre  suggested.  These  include  p  (persevera- 
tTon), "o  (oscillation) ,  and  w  (will),  the  last  extending  the 
theory  to  the  field  of  personality  traits.  J[t  has  also  been 
proposed  by  Spearman  (cf.  28)  that  whereas  i%^N  rep  repents 
tji£  total ,  amount  of  .mental  energy  at  the  subject's  dis- 


posal,  p  may  denote  the  inertia  of  such  mental  energy, 
-ag^^^i^rigtfeadiriess  of  its  supply.  Thus  all  three  gen- 
eral factors  may  be  but  different  manifestations  or  aspects 
of  ,  the  same  fundamental  factor. 

Spearman  at  present  recognizes  the  presence  of  all 
three  possible  classes  of  factors,  viz.,  general,  group,  and 
specific.  The  chief  differentiating  characteristic  of  his 
theory  ..seems,  to  be  its  emphasis  upon  the  g  factor  as,  the 
predominant  influence  in  correlation  and  its  relegation 
of  .grpup^factors  to  a  position  of  minor  importance. 

Thomson  (31-34  and  7)  has  proposed  a  Sampling  Theory 
of  mental  organization,  which  has  undergone"  'lifEleTor*  no 
change  since  its  original  formulation.  Behavior,  according 
to  Thomson,  depends  upon  a  very  large  number  of  inde- 
pendent elements  which  he  has  occasionally  identified 
with  neurones  or  bonds  between  neurones.  Any  one 
activity  of  the  individual  depends  upon  or  involves  a 
particular  sample  or  pattern  of  these  elements.  w  Correla- 
tion^resuTts  from  the  overlapping  of  different  samples,  of 
elements.  There  may  thus  be  found  any  number  of  dif- 
ferent types  of  factors,  varying  from  the  specific,  through 
group  factors  of  differing  extent,  to  the  general.  Thom- 
son has  repeatedly  illustrated,  with  data  from  dice  throws,1 
how  various  factor  patterns  could  result  from  overlapping 
samples  of  independent  elements. 

not  due  to  i 

elementary  abilities  involved  bu^jjjL^ 
aad«,diJ^  .abilities.    As  a  practical 

illustration  of  this,  Thomson  cites  the  well-known  drop- 
ping out  of  unnecessary  movements  in  the  learning  of  a 
motor  skill. 

1  Frequently  employed  in  statistics  as  a  means  of  obtaining  purely  random,  or 
"chance"  data. 


Thorndike's  views  on  the  relationships  of  mental  traits 
seemTo^have  run  the  gamut  from  extreme  specificity  to 
the  opposite  extreme  of  a  single  general  factor.  Thorn- 
dike's  belief  in  strict  specificity  is  ordinarily  traced  to  his 
1909  article  with  Lay  and  Dean  (40)  in  which,  after 
reporting  very  low  correlations  between  tests  of  sensory 
discrimination  and  estimates  of  intelligence,  Thorndike 
concluded:  "In  general  there  is  evidence  of  a  complex 
set  of  bonds  between  the  psychological  equivalents  of 
both  what  we  call  the  formal  side  of  thought  and  what 
we  call  its  content,  so  that  one  is  almost  tempted  to  re- 
place Spearman's  statement  by  the  equally  extravagant 
one  that  there  is  nothing  whatever  common  to  all  mental 
functions,  or  to  any  part  of  them"  (40,  p.  368). 

Similarly,  in  the  1914  edition  of  Thorndike' s  Educational 
^Psychology     (35)     appeared    the    following    statements: 
c£.  .  .  the  mind  must  be  regarded  not  as  a  functional 
ticjut,  nor  even  as  a  collection  of  a  few  general  faculties 
2  fiich    work    irrespective    of    particular    material,    but 
facther  as  a  multitude  of  functions,  each  of  which  involves 
fa jntent  as  well  as  form,  and  so  is  related  closely  to  only  a 
riew  of  its  fellows,  and  to  others  with  greater  and  greater 
degree  of  remoteness  ...  we  need  to  bear  in  mind  the 
singularity  and  relative  independence  of  every  mental 
process,  the  thoroughgoing  specialization  of  the  mind" 
(35?  PP-  366-367).   Thorndike  also  pointed  out  that  "the 
circumstances  of  training  would  seem  to  sometimes  in- 
tensify and  sometimes  weaken  original  relations"  (p.  371)- 
It  will  be  noted  that  the  possibility  of  narrow  group 
factors  is  admitted  even  in  these  early  statements,  and 
the  influence  of  environment  in  altering  the  organization 
of  abilities  is  suggested. 

In    1921,  following   the    analysis   of  intercorrelations 
among  the  subtests  of  Army  Alpha  and  Army  Beta, 


Thorndike  (37)  suggested  the  possibility  of  fairly  broad 
group  factors,  including  numerical  and  spatial  aptitudes 
among  others.  In  a  symposium  on  intelligence  and  its 
measurement  (38)  appearing  in  the  same  year,  he  stated: 
"  We  know  that,  taking  people  as  we  find  them,  the  ability 
measured  by  verbal  tests  is  not  the  same  as  the  ability 
measured  by  non-verbal  tests;  and  there  is  reason  to 
expect  other  similar  specializations"  (38,  p.  126).  But 
in  this  case  specialization  means  group  factors  of  fairly 
broad  extent.  At  the  same  time,  more  emphasis  is  shifted 
to  the  environment  as  an  influence  which  might  produce 
specialization.  This  is  illustrated  by  the  following  state- 
ment: "All  of  the  above,1  of  course,  concerns  individuals 
as  we  find  them,  products  of  nature  and  nurture.  Spear- 
man's doctrine  might  fit  the  original  nature  of  intellect 
better.  Certain  factors,  like  ability  to  understand  oral 
language,  ability  to  read,  ability  to  perceive  objects  in 
three  dimensions,  which  occur  to  everybody  as  neither 
entering  into  all  cognitive  performances  of  a  person  nor 
entering  into  only  a  few  very  closely  similar  performances, 
might  in  original  nature  be  absorbed  into  one  unitary 
ability  to  learn"  (38,  p.  151). 

whereas  "between^gg^ 


(36).  It  is  doubtful,  however,  whether  this  analysis 
should  be  incorporated  into  a  survey  of  Thorndike's  basic 
theory  since  it  was  offered  only  as  a  practical  suggestion 
to  expedite  testing.  In  any  event,  it  falls  in  line  with  the 
above  suggestions  of  broader  group  factors. 

In    The  Measurement  of  Intelligence  (41),  Thorndike 

1  Referring  to  specialization. 


swings  all  the  way  to  a  general  factor  theory  with  his 
Quantity  Hypothesis.  This^  hypothesis  "asserts  that  in 
their  deeper  nature  the  higher  forms  of  intellectual  opera- 
tion are  identical  with  mere  association  or  connection 
forming,  depending  ozpon  the  same  sort  of  physiological 
connection  but  requiring  many  more  of  them.  By  the  same 
argument  the  person  whose  intellect  is  greater  or  higher 
or  better  than  that  of  another  person  differs  from  him  in 
having  .  .  .  simply  a  larger  number  of  connections  of 
the  same  sort"  (41,  p.  415).  Thorndike  also  proposes 
that  "we  may  be  able  for  many  "purposes  to  replace  our 
measurement  via  a  sample  inventory  of  tasks,  by  a  more  or 
less  direct  measurement  of  C"  (41,  p.  422).  By  "C".he 
refers  to  the  total  number  of  connections  which  the  indi- 
vidual can  possibly  have  by  original  capacity.  Attention 
is  called  to  the  fact  that  this  hypothesis  limits  itself  to  the 
organization  of  original  capacity,  and  that  various  other 
relationships  may  be  environmentally  produced.  The  chief 
difference  between  the  Quantity  Hypothesis  and  Spear- 
man's theory  seems  to  be  that  in  the  former  the  number 
of  possible  connections  is  substituted  for  mental  energy 
as  the  interpretation  of  the  general  factor.1 

Lastly,  we  may  consider  what  can  conveniently  be 
classified  as  Group  Factor, uO&  Multiply  £<wtor  Theories. 
A  group  factor,  it  will  be  recalled,  is  one  which  is  common 
to  only  a  group  of  activities;  it  is  narrower  in  extent  than 
the  general  factor  and  broader  than  specifics.  The  publi- 
cation in  1928  of  K^iley's  Crossroads  in  the  Mind  of  Man 
(18)  paved  the  way  for  a  large  number  of  studies  in  quest 
of  particular  group  factors.  lili.dleya^ 
critical  analysis  of  the  methodology  and  data  of  Spear- 

1  Recently,  Thorndike  has  again  emphasized  specialization,  on  the  basis'  of  the 
current  findings  on  trait  relationships.  He  points  out,  for  instance,  that  many  of 
the  recently  proposed  group  factors  "correspond  interestingly"  to  "conceivable 
biological  realities"  (cf.  39). 


man,  that  the  general  factor  is  of  relatively  minor  im- 
portance and  can  usually  be  attributed  to  the  hetero- 
geneity 1  of  the  subjects  and  to  the  common  verbal  nature 
of  the  tests  employed.  If  a  residual  general  factor  be 
found  when  these  influences  are  ruled  out,  Kelley  claimed 
that  it  would  probably  be  very  small  and  insignificant. 
The  major  relationships  among  tests  he  attributed  to  a 
relatively  small  number  of  broad  group  factors.  Chief 
among  these  were  manipulation  of  spatial  relationships, 
facility  with  numbers,  facility  with  verbal  material, 
memory,  and  mental  speed.2 


It  is  not  within  the  scope  of  this  book  to  discuss  the 
mathematical  techniques  which  have  been  developed  and 
refined  for  the  analysis  of  trait  relationships.  This  mate- 
rial is  of  too  technical  a  nature  to  be  of  interest  to  the 
general  reader*  It  is,  however,  necessary  to  learn  some- 
thing of  the  general  approach  to  the  problem  and  of  its 
limitations.  The  present  section  is  intended  to  serve 
chiefly  as  an  orientation  for  the  proper  understanding  of 
the  results  to  be  subsequently  reported. 

Fundamentally,  all  techniques  for  the  study  of  mental 
organization  are  based  upon  the  correlation  coefficient. 
This  measure  indicates  the  degree  of  relationship  between 
two  sets  of  scores,  or  the  extent  to  which  each  individual's 
performance  in  one  test  corresponds  to  his  performance  in 
another  test.  Correlation,  however,  cannot  analyze  the 

1  The  influence  of  heterogeneity  upon  correlation  coefficients  will  be  discussed 
in  the  following  section. 

2  Various  modifications  of  group  factor  theories  have  appeared.    Thurstone 
(42)  identifies  himself  with  some  form  of  group  factor  theory,  although  he  has 
been  more  concerned  with  the  development  of  methods  than  with  the  formula- 
tion of  a  definite  interpretation.  Variations  of  group  factor  theory  have  also  been 
proposed  by  Garrett  (14),  Hull  (16),  Meili  (20),  and  Tryon  (43).    For  a  fuller 
description  of  each,  see  Anastasi  (5),  Ch.  I. 


mutual  interrelationships  of  a  large  number  of  variables. 
A  correlation  coefficient  may  indicate  whether  there  is 
some  factor  common  to  a  pair  of  tests,  but  it  cannot  show 
the  presence  of  a  single  common  factor  through  three,  or 
four,  or  any  larger  number  of  tests.  Let  us  suppose  that 
all  intercorrelations  among  three  tests  have  been  com- 
puted, with  the  following  results:  1 

r  12  =  .60 
fis  =  49 

7*23    =    -7° 

Although  all  three  correlations  are  positive  and  high,  we 
cannot  determine  whether  these  three  tests  have  one  com- 
mon factor  or  several  common  factors  among  them. 
Test  i  might  share  one  factor  A  with  test  2,  and  a  differ- 
ent factor  B  with  test  3  ;  a  still  different  factor  C  might 
constitute  the  common  element  between  tests  2  and  3  . 
It  was  Spearman  (26)  who,  first  demonstrated  that 

coefficients  it  is 


traits.    TheUin:^^  Spearman  was  the 

hierarchal  arrangement  of  correlation  coefficients.  Accord- 
ing to  this  criterion,  if  it  was  possible  to  arrange  all  the 
ifftercorrctetiSris  among  a  set  of  tests  in  such  a  way  that 
they  decreased  consistently  in  size  both  aloag  .the  rows 
anH^Iong  tfiFcoTumns  oFtEe  Stable,  then  the  relationships 
an^  ,  entirely,  in,  terms  of 

gjtnd^.  This  was  a  relatively  crude  "inspectional"  method 
of  determining  hierarchy.  Subsequently^  t^e  computa- 
tion of  intercolumnar  correl^L^fT^w^B  suggested  as  a 

p.^  ^-"""'"""'^^  I(MJ  ((9!lf  ,«,,«.  i»      •'"«<  "     ',.,     ,    OQ'i 

conjsiiifiiik,,  numenc*ff^tfex  of  hierarchy.  The  inter- 
columnar  correlation1"*  is  simply  the  correlation  between 
columns  of  correlation  coefficients.  A  +  i.oo  intercolum- 

1  It  is  customary  to  denote  the  particular  variables  correlated  by  subscripts. 
Thus,  ri2  is  the  correlation  between  test  No.  I  and  test  No.  2. 


nar  correlation  would  indicate  a  perfect  hierarchal  arrange- 
ment of  the  coefficients. 

Fifially,.Jthje.mtercoluin-nar  correlation  was  replaced  by 
thej£^i,ttrx^tip^.  The  latter  gets  its  name  from  the 
fact  that  the  tests  are  considered  in  sets  of  four.  For  every 
four  tests,  or  variables,  we  can  compute  three  tetrad  equa- 
tions as  follows: 

^1234  =  7*12  X  7*34  ~  ?*13  X  7*24 
^1243  =  7*12  X  7*34  —  7*14  X  T^ 
^1342  =  7*13  X  7*24  —  7*14  X  7"23 

?rs  ,have  been,  able,™  to  prove  ~  .tn^th^ 
.three  tetrad  equations  are  equal  ;to 
single  common  factor  Is  sufficient  to  account 

d^^^  the  four  varkbles. 

This  was  a  decided  step  forward  from  the  simple  cor- 
relation coefficient.  It  was  now  possible  to  analyze  the 
interrelationships  of  any  number  of  variables  by  comput- 
ing different  sets  of  tetrads.  The  extension  of  the  tetrad 
criterion  beyond  four  variables  can  be  easily  demon- 
strated. Let  us  suppose  that  we  have  administered  six 
tests  to  the  same  subjects.  First,  we  compute  the  three 
tetrad  equations  with  tests  I,  2,  3,  and  4.  If  all  three  tet- 
rads are  equal  to  zero,  we  may  conclude  that  the  same 
factor  which  underlies  tests  I  and  2  is  also  common  to 
tests  3  and  4.  Then  if  the  tetrad  criterion  is  likewise 
satisfied  (i.e.,  all  tetrads  equal  to  zero)  with  tests  i,  2,  5, 
and  6,  we  know  that  the  factor  common  to  I  and  2  is 
identical  with  that  common  to  5  and  6.  Hence  the  same 
factor  must  be  common  to  all  six  variables. 

^^  ^k^JlE^ 

remained  on 

1  Within  their  probable  errors. 


speculation.  In  its  specific  application  and  interpretation, 
however,  the  tetrad  criterion  has  been  the  target  of  much 
criticism.  Its  chief  critic  has  probably  been  Thomson. 
The  latter  has  repeatedly  pointed  out  and  demonstrated 
that  the  tetrad  criterion  can  be  satisfied  without  a  g  factor 
(cf.  31,  32,  33).  The  .satisfaction  of  the  tetrad  criterion 
merely  indicates  that  the  observed  relationships  among 
the  given  variables  could  be  explained  in  terms  of  g  and  s. 
But  they  could  also  be  accounted  for  by  other  factor 
patterns.  In  other  words,  tetrad  equations  can  never  dis- 
prove the  possibility  of  alternative  factor  analyses  of  the 

The  tetrad  criterion  is  also  inadequate  in  that  it  does 
not  in  itself  indicate  the  magnitude  or  relative  importance 
of  the  common  factor  in  each  variable.  Similarly,,  when 
the  presence  of  group  factors  is  demonstrated,1  the  weights 
of  these  factors  in  the  various  tests  cannot  be  determined. 
Recently,  more  elaborate  and  exact  procedures  have  been 
evolved  for  the  direct  analysis  of  any  factor  pattern. 
These  methods  have  become  known  under  the  name  of 
factor  pattern  analysis*  Like^tetrads,  they  are  ultimately 
Based  upon  certain  relationships  among  correlation  co- 
efficients. It  is  possible  by,  these, juelJiada^tg^arnve  at 
thL ".weight"  or  "loading"  of ^aph,,f actor  jn^each  of  the 
^ap&bles.  A  sample  of  such  a  factor  pattern  analysis  is 
shown  below. 

In  this  factor  pattern,  the  numbers  below  each  factor 
show  the  weight  of  that  factor  in  each  test,  or  the  degree 
to  which  performance  in  the  test  is  attributable  to  that 
particular  factor.  Thus  factor  I  has  a  large  and  positive 
weight  in  all  four  tests  and  is  therefore  a  general  factor 

1  I.e.,  when  the  tetrad  equations  are  not  equal  to  zero. 

2  Several  different  methods  of  factor  analysis  have  been  worked  out;  cf.,  e.g., 
Hotelling  (15),  Kelley  (19),  Thurstone  (42). 




(After  Hotelling,  15,  p.  434) 


Factor  I 

Factor  II 

Factor  III 

Factor  IV 

Reading  speed 



—  .292 


Reading  power 


—  »62O 



Arithmetic  speed 





Arithmetic  power 





Percentage   of   total   var- 

iance    attributable      to 

each  factor 





common  to  all  of  these  tests.  Factor  II  reveals  a  differ- 
entiation between  verbal  and  arithmetic  ability,  since 
it  has  positive  weights  in  the  two  arithmetic  tests  and 
negative  weights  in  the  verbal  ones.  Factor  III  has 
positive  weights  in  the  two  power  tests  and  negative 
weights  in  the  two  speed  tests,  thus  suggesting  a  possible 
distinction  between  speed  and  deliberation,  or  careful- 
ness. The  fourth  factor  is  too  small  to  be  of  any  signifi- 
cance. The  percentages  given  in  the  last  row  of  the  table 
indicate  the  relative  importance  of  the  four  factors  in  all 
of  the  tests.  Thus  the  first  factor  alone  is  sufficient  to 
account  for  nearly  one-half  (46^%)  of  the  relationships 
found  among  these  four  tests,  the  second  factor  for  36!%, 
the  third  for  13%,  and  the  fourth  for  a  negligible  4%. 

It  should  be  kept  in  mind,  that  the  same  sort  of  gualifi- 
catioris  made  in  the  interpretation  of  tetrads  applies  to 
fSS^T'p^i^^^^^y^^''  In  ho  case  is  the  possiDmty  of 
alternative  explanations  precluded.  As  Thomson  has 

pointed  out,  with  Tny  given  body  of ^  data, ^^JJjn 


twS^"^^  ,.on .  psy^hjplggi^l,wS"ounds" 

(34,  p.  185).   Any  set  of  intercorrelations  can  be  analyzed 


into  factors  in  an  infinite"  number  of  ways.  In  order  to 
arrive  at  a  determinate  solution,  certain  "limiting  condi- 
tions" must  be  imposed.  The  various  current  methods 
of  factor  analysis  differ  in  their  choice  of  limiting  condi- 
tions^ prpbitulates.  It  might  be  noted,  however,  that  in 
actual, practice  the  general  results  obtained  by  these 
djjfexeJiLmethods  do  not  differ  very  significaiitly. 

In  concluding  this  very  brief  survey  of  methodological 
problems,  mention  should  be  made  of  the  widely  discussed 
question  ofjieterogeneiity.  It  has  been  repeatedly  demon- 
strated, both  empirically  (cf.  10)  and  theoretically  (cf.  13), 
that  the  size  of  a  correlation  coefficient  is  affected  by  the 
heterogeneity  of  the  group  of  subjects  upon  whom  the 
data  were  collected.  The  most  obvious  example  is  that 
of  age  heterogeneity.  If  the  subjects  range  in  age  from 
3  to  15  years,  a  high  positive  correlation  will  be  found 
between  even  such  diverse  characteristics  as  size  of  the 
great  toe  and  Stanford-Binet  mental  age.  The  same  two 
measures  would  yield  a  zero  correlation  within  a  homo- 
geneous age  group  such  as,  for  example,  lo-year-old  chil- 
dren. N^does?  festewgeae^ 

tio^  coefficient.  If  a  heterogeneous  group  composed  of 
Chinese  and  Scandinavians  were  rated  for  height  and  for 
proficiency  in  the  use  of  chopsticks,  a  fairly  high  negative 
correlation  would  be  obtained  between  these  two  measures. 
The  Chinese  subjects  would,  in  general,  be  shorter  than 
the  Scandinavians  and  definitely  more  adroit  with  chop- 
sticks. Within  either  group,  however,  we  should  scarcely 
expect  any  correlation  between  these  two  characteristics. 

w;w^^  fcom-  a  -^narked degree   of 


It  is  difficult  to  decide,  however,  just  what 
constitutes  a  permissible  degree  of  heterogeneity.  Obvi- 
ously, all  heterogeneity  should  not  be  eliminated,  even 


if  this  were  possible,  since  individual  differences  would 
thereby  disappear  and  correlation  would  be  meaningless. 
The  desired  degree  of  heterpg^eity:,,tmust  be  determined 
arbitrarily  and  on  the  basis  of  the  particular  problem 
under  investigation.  It  should  always  be  remembered, 
hpj5^verjwthat  a  correlation  coefficient,  or  any  statistical 
measure  derived  therefrom,  must  be  interpreted  in  th,e 
t^  of  the  particular  group  upon  which  it  was  obtained. 

Perhaps  the  most  general  finding  of  the  numerous  and 

on  mental  organization  is  the  pjes- 
c£j)i  group  factors  of  varying  extent.   ^wiUJpe  recalled 
that  „  all  theories  now  admit  the  presence,  of  such  factors, 
differing  in  the  relative-  emphasis  which  ,th^y 
em.  There  is  scarcely  a  single  well  controlled 

study  which  has  not  revealed  group  factors  of  some  sort. 
The  in^  he  ,organ- 

.relatively  ,sm4LnumKez  j?H^^ 

may  copibiae  in  v-ar,iou^^w-ay&^m-aay.,,,Qne 

Of  the  group  factors  proposed  by  Kelley  (18),  definite 
evidence  has  been  found  for  the  verbal  and 

factors  (cf.,  e.g.,  21,  24,  29).  Sp^ial  aptitude  seems  to*be 
a  composite  of  more  than  one  indepe^4wt^£actor.  The 
administfatiofi  "of  tEe'extensive  series  of  Minnesota  Me- 
chanical Aptitude  Tests  to  groups  of  high  school  boys 
revealed  a  number  of  narrow  group  factors  rather  than 
one  unitary  factor  in  the  ability  to  deal  with  spatial  rela- 
tions (22).  Other  investigations  have  indicated  the  pres- 
ence of  "  practical  intelligence  "  (i)  which  is  predominant  in 
.the  performance  type  of  intelligence  test.  In  addition,  inde- 
pendent factors  of  "  mechanical  aptitude  "  (8)  and  of  "  rou- 
tine manual  aptitude"  (9)  have  been  tentatively  established. 


The  latter  is  found  among  motor  skills  and  manipulatory 

In  regard  to  a  possible  common  factor  of  speed,  the 
data  are  very  conflicting.  In  general,  there  seems  little 
basis  for  expecting  an  individual  who  is  fast  in  one  sort 
of  mental  activity  to  be  proportionately  fast  in  others. 
Speed  of  performance  seems  to  be  quite  highly  specific 
and  to  depend  upon  the  nature  of  each  particular  task. 
When  evidence  is  presented  for  the  existence  of  an  inde- 
pendent group  factor  (cf.,  e.g.,  11),  such  a  factor  is  very 
small  and  is  cut  across  by  numerous  other  group  factors. 

Thq  ,iame  is  true  of  memory.  Individuals  cannot  be 
characterized  as  possessing  a  good  memory  or  a  poor 
memory,  since  they  do  not  manifest  uniform  .powers  of 
retention  for  different  materials.  A  common  factor  through 
memory  tests  is  found  only  when  the  tests  are  also  similar 
in  other  respects.  For  example,  if  all  are  rote  memory 
tests  for  relatively  meaningless  material  and  all  the  items 
are  presented  visually,  then  performance  on  such  tests 
will  show  a  common  factor  (cf.  2).  In  such  cases,  however, 
the  same  special  devices  or  aids  to  memorization  can  be 
applied  to  all  the  tests.  Thus  if  a  subject  evolves  the  de- 
vice of  forcing  a  meaningful  association  into  the  material, 
this  will  help  him  on  all  the  tests  and  thereby  produce  a 
certain  uniformity  of  performance.  But  such  is  not  the 
case  when  memory  is  tested  in  a  variety  of  situations, 
including  recall  and  recognition  of  logical  material,  rote 
memory  for  numbers  and  for  letter  combinations,  memory 
for  tones,  etc.  (cf.  3). 

In  the  measurement  of  personality,  the  same  techniques 
of  factor  pattern  analysis  are  being  gradually  introduced. 
Thus,  for  example,  among  the  various  items  included  in 
the  Bernreuter  Personality  Inventory,  two  independent 
group  factors,  or  "traits"  have  been  identified  (12).  These 


have  been  tentatively  described  as  sociability  and  self- 
confidence.  Various  investigators  have  undertaken  factor 
pattern  analyses  of  emotional  stability,  interests,  and 
other  phenomena  of  personality.  Conclusions  in  regard 
to  the  organization  of  personality  traits  are  still  highly 
tentative,  however,  owing  to  the  complex  nature  of  the 
phenomena  under  investigation  and  to  the  relatively  recent 
application  of  testing  techniques  to  this  field. 


The  reader  may  have  noticed   that  those  aptitudes 
whidGTave^Been  jhpwn  £0  be  Jndep  creator;  differentiable 

by  community,  of  mate^M 

content.  Wh^njthe  similarity  among  the  tests  was  onejof 
"process,  structure,  orjorm,  no  clear-cut  group  factor  could 
^^^^.^^^^^  them.  Such^  was  the  case  with  speed 
and  memory.  Both  may  be  tes  ted  .  with  any  kind  of  ma- 

.  .  ^ 

and  numerical  factors  are  definitely  linked,,,  up  with  m^- 
^^^^^^^^^  factor  will  be  found  through  verbal 
tests,  regardless  of  what  the  subject  is  required  to  do 
with  the  words.  The  individual  who  ranks  relatively 
high  in  verbal  aptitude  will  excel  in  verbal  memory,  will 
be  faster  on  a  verbal  test,  more  adept  at  perceiving  rela- 
tions which  are  expressed  linguistically,  and  so  on.  Thus 
content  factors  cut  across  the  boundaries  of  form  and 
seem  to  be  the  most  potent  determiners  of  uniformity  of 

Direct  evidence  for  this  conclusion  can  be  found  in 

1  No  fundamental  differentiation  between  form  and  content  is  implied  by  this. 
The  distinction  is  simply  a  convenient  practical  one.  To  be  sure,  a  group  factor 
is  simply  an  index  of  relationship  among  the  individual's  responses  to  concrete 
stimuli.  Similarity  in  that  aspect  of  the  situation  which  is  commonly  designated 
as  "material"  or  "content"  seems  to  exert  more  influence  upon  such  relation- 
ships, however,  than  similarity  in  any  other  aspect. 


several  studies  which  have  employed  tests  differing  both 
in  form  and  in  content.  In  an  investigation  with  a  wide 
variety  of  memory  tests  (3),  for  example,  the  specific 
materials  of  the  tests  were  found  to  be  much  more  effective 
in  determining  correlation  than  the  methods  of  testing 
memory.  Thus  a  logical  recall  and  a  logical  recognition 
test,  both  involving  prose  passages,  correlated  .74  with 
each  other;  it  should  be  noted  that  these  two  methods  of 
testing  memory  are  generally  considered  the  most  unlike. 
Similarly,  the  logical  recall  and  recognition  tests  correlated 
.42  and  .56,  respectively,  with  a  verbal  analogies  test. 
The  correlations  of  these  two  memory  tests  with  other 
memory  tests,  on  the  other  hand,  were  much  lower. 
Logical  recognition  correlated  only  ,23  with  nonsense 
syllable  recognition,  for  example,  although  both  tests 
employed  the  same  method  of  testing  memory. 

In  an  investigation  designed  specifically  to  test  the 
relative  influence  of  " material"  and  "structure"  in  men- 
tal organization  (25),  more  conclusive  evidence  for  the 
same  point  was  obtained.  A  series  of  14  tests  was  admin- 
istered to  1 86  male  college  students.  In  respect  to  material, 
four  of  the  tests  were  numerical,  four  spatial,  and  five 
verbal.  At  least  three  "structural  patterns"  were  repre- 
sented, viz.,  analogies,  generalizations,  and  construction;  * 
each  of  these  occurred  in  all  three. types  of  material.  A 
factor  pattern  analysis  of  the -intercorrelations  .among 
these  tests  indicated;  that  .similarity  of  material  was  more 
influential  than  similarity  of  structure  in  producing -group 
factors.; ;  .  ..'.*."  ' .  \''.  ':.''" 

It  is  inter££tW3guto  compare  tlie  empirically  discovered 
group* factors  with  the  proposed, mental  " faculties"  of  the 
medieval  Scholastic  philosophers,  which,  have  found  their 

1  Including  such  tests  as  sentence  completion  and  number  series  completion, 
as  well  as  spatial  construction  tests. 


way  Into  popular  speech.  Whereas  the  "  faculty  psycholo- 
Ssts^  spoke  oFatteatipn,  mSnory3  ;  judgment,  reasoi^jng, 
ancTsTmilar  mental  processes,  we  now  find  group  factors  of 
verEaT,  "numerical,  and  spatial  aptitudes.  The  latter  clas- 
sification Is  one  of  content-  the  Scholastic  was  one  of 


With  the  rapid  accumulation  of  data  on  mental  organ- 
ization, certain  definite  discrepancies  have  been  brought 
to  light.  Rather  than  being  stable  and  universal,  trait 
'  beeu  fpuad  tp.,vaj:y  fcgin  jpne^type  of 
Sgearman  called  attention  to  these 

sjIB|ec?tgla|  Sgearman  called  attention  to  these 
differences  in  1927,  stating  *that,  "Another  important 
influence  upon^he  saturation  of  an,  a^Uity  with.  g-  Appears 
to  be,  the  class  of  person  at  issue"  (27,  p.  217).  At  this 
time,  he  also  reported  some  data  which  suggested  that 
among  older  as  well  as  among  brighter  individuals,  abilities 
were  more  specialized  and  the  general  factor  played  a 
relatively  less  dominant  part. 

Although  the  data  of  earlier  studies  are  somewhat 
conflicting,1  recent  and  more  adequately  controlled  in- 
vestigations have  demonstrated  a  tendency  for  abilities 
to  be  more  s^cm^e^^mori^  Bolder  ^^^^j^ttf^mqp.g 
thqse  'i&skJ%kf*  educational  -levels.-  TEus  investigations  on 
schoolchi^  the  presence  of^an 

appreciab^  (cf.,  e.g.,  23).  It  is  interesting 

*  'Jiaycs^mb^  of  the  (  studies  ,  J>y^  Sjaearman^ 

.CC^iaB^I.-PR^  ^ 


Studies  on  college  groups  (cf.,  e.g.,  2,  3,  24),  on  the  other 
hand,  have  indicated  a  much  more  pronounced  specificity, 

1  Cf  .  Anastasi  (5)  for  a  discussion  of  these  data. 


with  relatively  narrow  group  factors.  Recently ,  ^  some 
suggestive  data  have  also  been  presented  to  show  diverse 
factor  patterns  among  various  occupational  groups  (cf.  20). 
Certain  interesting  correspondences  were  noted  between 
the  nature  of  the  occupational  activities  and  the  organiza- 
tion of  the  individual's  abilities. 

Although  these  discrepancies  originally  led  to  contro- 
versy or  confusion,  they  are  now  coming  to  be  recognized 
as  an  important  means  of  studying  the  nature  of  trait 
relationships.  Thus  two  recent  studies  have  been  con- 
ducted with  the  explicit  purpose  of  discovering  age  changes 
in  mental  organization.  In  one  of  these  (14),  three  groups 
of  schoolchildren,  aged  9,  12,  and  15  years,  respectively, 
were  given  six  memory  tests  as  well  as  tests  of  motor 
speed  and  verbal,  numerical,  and  spatial  aptitudes.  The 
intercorrelations  among  these  tests  tended  to  decrease 
with  age,  as  is  indicated  by  the  average  intercorrelations 
of  .29,  .26,  and  .14  obtained  in  the  9,  12,  and  15  year 
groups,  respectively.  Factpjq  p,a£tern analyses  revealed  a 
large  general  factor  whose  weight  decreased,  consistently 
from  ages  9  to  15. 

In  the  secondjtudy  (6),  a  single  group  ,Q£  .children  were 
retested  after  a  lapse  of  three  years.1  Eight  tests,  includ- 
ing verbal,  numerical,  and  spatial  materials,  were  admin- 
istered. ^Intercorrelations  dropped  from  the  first  to  the 
second  testing^  this  decrease  being  greater  in  the  correla- 
tions between  verbal  and  numerical  tests  than  in  those 
within  either  group.  Factor  pattern  analyses  corroborated 
these  findings.  At  both  age  levels,  a  large  general  factor 
was  found,  but  its  magnitud^decteasjedx^ 
9  to  12. 

'**H.l,  n     ,)    ,  I    ><  '      -fK 

xThe  initial  data  were  collected  in  an  earlier  investigation  (23)  on  mental 
organization  with  a  group  of  395  children.  Of  the  latter,  it  was  possible  to  obtain 
161  for  retesting  three  years  later. 


In  both  of  these  studies,  the  changes  in  mental  organiza- 
tion may  have  resulted  from  a  variety  of  conditions. 
The  subjects  differed  in  age,  in  general  experience  and 
training,  and  in  academic  level.  It  is  therefore  .impossible 
to  determine  what  brought  about  the  change  in  mental 
organization.  It  might  be  argued  that  mental  organiza- 
tion undergoes  a  progressive  and  regular  alteration  with 
age.  On  the  other  hand,  the  variations  observed  may  have 
resulted  from  environmental  stimulation  and  in  partic- 
ular from  school  instruction.  The  question  can  only  be 
answered  by  an  experimental  approach.  If  the  pattern  of 
trait  relationships  can  be  experimentally  altered,  the  in- 
fluence of  environmental  stimulation  upon  mental  organ- 
ization will  have  been  demonstrated. 

This  was  undertaken  in  a  recent  experiment  by  the 
writer  (5).  The  essential  aim  of  the  study  was  the  altera- 
tion of  a  factor  pattern  by  a  brief,  relevant,  interpolated 
experience.  Five  tests,  including  vocabulary,  memory 
span  for  digits,  verbal  reasoning  of  the  syllogistic  type, 
code  multiplication,  and  pattern  analysis,1  were  admin- 
istered to  200  sixth  grade  schoolchildren.  All  subjects 
were  then  given  instruction  in  the  use  of  special  tech- 
niques or  devices  which  would  facilitate  performance  on 
the  last  three  tests  only.  This  instruction  was  similar 
in  its  general  nature  to  that  received  in  the  course  of 
school  work,  as,  for  example,  in  the  teaching  of  arithmetic 
operations,  short-cuts  of  computation,  etc.  After  a  lapse 
of  13  days,  parallel  forms  of  all  five  tests  were  administered 
under  exactly  the  same  conditions  as  in  the  initial  testing. 

age  changes  were  reduce^ 

Ini^^SI^^i^IA  janifbxm^ 
ves^g^itionon  this  problem. 

1  Predominantly  a  spatial  ability  test. 


For  convenience  in  discussing  the  data,  we  shall  refer 
to  vocabulary  and  memory  span,  the  two  tests  in  which 
no  instruction  was  received,  as  "non-instruction"  tests; 
the  remaining  three  tests  will  be  referred  to  as  "instruc- 
tion" tests.  Ajzomgarison  of  the  intercorrelations  among 
the  five  variables  in  the  initial  and  final  testing  showed 
practically  no  change  in  the  correlation  between  the  two 
"non-instruction"  tests,  a  slight  change  in  the  correla- 
tions between  "instruction"  and  "non-instruction"  tests, 
a.nd  a  marked  change  in  the  correlations  among  the  three 
"instruction"  tests.  Factor  pattern  analyses  l  revealed 
a  wide  variation  from  the  initial  to  the  final  testing.  An 
examination  of  the  weights  of  each  factor  in  the  five 
tests  before  and  after  instruction  suggested  that  the 
changes  were  such  as  would  have  been  expected  from  the 
nature  of  the  interpolated  experience. 

All  of  the  above  investigations  indicate,.  that-"  factors" 
cannot  be  regarded  as  fixed  and  immutable.  Nor  can 
they  be  interpreted  in  terms  of  underlying  psychological 
entities.  A  "factor"  is  simply  a  mathematical  statement 
of  observed  relationships  among  a  group  of  concrete  be- 
havior manifestations  (cf.  4).  And  by  relationship  is 
meant  nothing  more  than  a  tendency  to  concomitant 
variation.  When  we  say  that  performance  on  two  tasks 
is  related,  we  mean  simply  that  those  individuals  who 
excel  in  one  tend  also  to  excel  in  the  other.  To  conclude 
any  more  than  this  from  a  factor  pattern  analysis  is  mere 


The  experiment  described  in 

1  Computed  by  Hotelling's  (15)  method  of  principal  components. 


the  preceding  section  reproduced,  in  a  highly  condensed 
form,  the  type  of  experience  to  which  the  child  is  exposed 
in  the  course  of  school  work  and  other  everyday  activities. 
It  is  therefore  entirely  possible  that  factor  patterns  are 
deteTmmecHn  the  first  place  by  the  nature  of  such  exp^ri- 
divers  e^J  actgr^pattemgL  .which  .  rhave  ,„  been 

found  in  differjerit,  populations,  nifty,  thus.,,  have,,  been  ,<BJQ- 
vironmental],)L4H:Qduced.  Insofar  as  the  experiences  of 
all  of  the  groups  employed  in  such  studies  have  common 
features,  certain  fundamental  similarities  in  their  mental 
organization  have  been  discovered.  Insofar,  however, 
as  these  experiences  have  varied  from  group  to  group, 
different  patterns  of  mental  organization  were  found. 
Thus  a  verj  widespread,  ..."yedxal...  factor!',  could  _  easily 
^in  'our  culture  in.  which,  language  plays  so  im- 
^rt'  in  a  wide  variety  of  fields.  If  there  were 
n'aT  system  in  which  only  woodcraft  and  poetry 
were  taught,  we  might  indeed  find  a  "  group  factor" 
common  to  these  two  abilities  and  to  no  other.  Those 
subjects  who  had  spent  more  time  and  effort  on  their 
academic  work  would  excel  in  both  tasks,  but  not  neces- 
sarily in  other  fields  of  activity.  The  experimental  find- 
ings which  have  been  reported  furnish  an  illustration  of 
how  "  group  factors"  might  be  produced  or  obliterated 
by  the  subject's  experience. 

It  also  iollowsjt^  may  vary  with 

agV^SEm^e^g,^^^^^^^8  •  "&s"  SK^^bjWt^s^x^ri- 
SnceTTccMtiuiate  throughout  his  life,  certain  alterations 
in  the  pattern  of  mental  organization  will  take  place. 
The  relationships  of  performance  in  the  same  or  similar 
situations  will  not  remain  constant  over  a  long  period  of 
time.  If  the  lapse  of  time  is  short,  the  influence  of  the 
intervening  experience  may  be  too  slight  to  produce  an 
appreciable  difference  in  factor  pattern.  Over  a  long 


period,  however,  the  cumulative  effect  will  become  ap- 
parent, as  in  the  comparison  of  children  with  adults. 

In  summary,  it  would  seem  that  the  relationships 
amoi^'^Tndividuars  performance  on  a  number  of  tests 
at  any  one  time  may  be  explained  in  terms  of  a  small 
number  of  independent  unitary  factors*  Under  existing 
cultural  "conditions,  a  certain  degree  of  uniformity  of 
factoif  pattern's  is  found  because  of  a  general  environ- 
mental uniformity.  Traditional  educational  curricula  and 
vocational  classifications  have  probably  contributed  much 
towards  this  uniformity.  Thus  in  the  young  schoolchild, 
we  find  a  large  general  factor  through  all  types  of  activities 
which  are  taught  in  our  schools,  the  so-called  "higher  cen- 
tal processes."  As  the  child  grows  older  and  specialization 
of  function  is  encouraged,  certain  culturally  determined 
differentiations  appear.  "Qrcmp  factors"  are  produced 

w-nuu^u. *-"«*'•»>  «-    '      '»" ' '         ,     <      .      "f"    *  ,   n  ,  ^"""1»»a«««»^^  «Jfc«'Jtt 

in  linguistic,  mathematical,  mechanical,  and  possibly 
other  functions.  These  f  actors,, .,  however,,  are  only  a 
mathematical  statement  or  conceptual  simplification  of 
the  observed  relations  among  concrete  responses.  And 
as  such  they  may  be  expected  to  shift  from  time  to  time 
in  the  same  subjects  or  from  one  population  to  another 
because  of  varying  experiences. 

Thus  the  role  of  environment  is  again  forcefully  demon- 
strated. In  view  of  the  dependence  of  the  individual's 
behavioral  development  upon  environmental  stimulation, 
it  is  not  surprising  to  find  that  the  relationships  among 
various  classes  of  behavior  are  also  environmentally  de- 
termined .  Thejais^^  of 
trait  relationship  could ^arcely,,,^ 
^traffr tKeniselves "have  been  found ,^tp,.  Jbo^Sfi 
influenced  by  experiential 



1.  Alexander,  W.  P.    "Intelligence,  Concrete  and  Abstract:  a 
Study  in  Differential  Traits,"  Brit.  J.  PsychoL,  Mon.  SuppL, 
No.  19,  1935.  Pp.  177. 

2.  Anastasi,  A.     "A  Group  Factor  in  Immediate  Memory," 
Arch.  PsychoL^  No.  120,  1930.  Pp.  61. 

3- "Further  Studies  on  the  Memory  Factor,"  Arch. 

PsychoL,  No.  142,  1932.   Pp.  60, 

4- "Some  Ambiguous  Concepts  in  the  Field  of  Mental 

Organization/'  Amer.  J.  PsychoL,  1935,  47,  508-511. 

5- "The  Influence  of  Specific  Experience  upon  Mental  Or- 
ganization," Genet.  PsychoL  Mon.,  1936,  18,  No.  4,  245-355. 

6.  Asch,  S.  E.   "A  Study  of  Change  in  Mental  Organization," 
Arch.  PsychoL,  No.  195,  1936.  Pp.  30. 

7.  Brown,  W.,  and  Thomson,  G.  H.    The  Essentials  of  Mental 
Measurement.      London:  ^Cambridge    Univ.    Press,    1925. 
Pp.  224. 

8.  Cox,  J.  W.  Mechanical  Aptitude — Its  Existence,  Nature,  and 
Measurement.  London:  Methuen,  1928.   Pp.  209. 

gf  t      Manual  Skill — Its  Organization  and  Development. 

London:  Cambridge  Univ.  Press,  1934.  Pp.  247. 

10.  Cureton,  E.  E.,  and  Dunlap,  J.  W.  "Some  Effects  of  Heter- 
ogeneity on  the  Theory  of  Factors,"  Amer.  /.  PsychoL, 
1930,  42,  608-620. 

11.  DuBois,  P.  H.    "A  Speed  Factor  in  Mental  Tests,"  Arch. 
PsychoL,  No.  141,  1932.  Pp.  38. 

12.  Flanagan,  J.  C,  Factor  Analysis  in  the  Study  of  Personality. 
Stanford  Univ.,  Calif.:  Stanford  Univ.  Press,  1935.  Pp.  103. 

13.  Garrett,  H.  E.,  and  Anastasi,  A.    "The  Tetrad-Difference 
Criterion  and  the  Measurement  of  Mental  Traits,"  Ann. 
N.  Y.  Acad.  Sri.,  1932,  33,  234-281. 

14.  Garrett,  H.  E.,  Bryan,  A.  I.,  and  Perl,  R.  E.    "The  Age 
Factor  in  Mental  Organization,"*  Arch.  PsychoL,  No.  176, 

I93S-    Pp- 31- 

15.  Hotelling,  H.    "Analysis  of  a  Complex  of  Statistical  Vari- 
ables into  Principal  Components,"  /.  Educ.  PsychoL,  1933, 
*4,  417-441; 


1 6.  Hull,  C.  L.    Aptitude  Testing.    Yonkers-on-Hudson:  World 
Book  Co.,  1928.  Pp.  535. 

17.  Kelleyv  T.  L.    Interpretation  of  Educational  Measurements. 
Yonkers-on-Hudson:  World  Book  Co.,  1927.  Pp.  363. 

1 8.  .   Crossroads  in  the  Mind  of  Man:  a  Study  of  Differ enti- 

able  Mental   Abilities.      Stanford   Univ.,    Calif.:   Stanford 
Univ.  Press,  1928.  Pp.  238. 

19. .  Essential  Traits  of  Mental  Life.   Cambridge:  Harvard 

Univ.  Press,  1935.   Pp.  145. 

20.  Meili,  R.    "Recherches  sur  les  formes  d'intelligence,"  Arch, 
de  psychol.,  1930,  22,  201-284. 

21.  N.  Y.  College  Entrance  Examination  Board:  Reports  of  the 
Commission  on  Scholastic  Aptitude  Tests ,  1926-1936. 

22.  Paterson,  D.  G.,  et  al.  Minnesota  Mechanical  Ability  Tests. 
Minneapolis:  Univ.  Minn.  Press,  1930.   Pp.  586. 

23.  Schiller,  B.     "Verbal,  Numerical,  and  Spatial  Abilities  of 
Young  Children,"  Arch.  Psychol^  No.  161,  1934.  Pp.  69. 

24.  Schneck,  M.  M.  R.     "The  Measurement  of  Verbal  and 
Numerical  Abilities,"  Arch.  Psychol.^  No.  107,  1929.   Pp.  49. 

25.  Smith,  G.  M.    "Group  Factors  in  Mental  Tests  Similar  in 
Material  or  in  Structure,"  Arch.  Psychol^  No*  156,  1933. 
Pp.  56. 

26.  Spearman,  C.    "  c  General  Intelligence  '  Objectively  Deter- 
mined and  Measured,"  Amer.J.  Psychol^  1904,  15,  201-293. 

27.  .    The  Abilities  of  Man.  N.  Y.:  Macmillan,  1927.  'Pp. 


28.  .   "  CG'  and  After— a  School  to  End  Schools."  In  Psy- 
chologies o/ipjo,  C.  Murchison,  ed.  Worcester:  Clark  Univ. 
Press,  1930.     Pp.  339-366. 

29.  Stephenson,  W.     "Tetrad-Differences  for  Verbal  Subtests 
Relative  to  Non-Verbal  Subtests,"  /.  Educ.  Psychol. ,  1931, 
22,  334-350- 

30.  Terman,  L.  M.    "The  Vocabulary  Test  as  a  Measure  of 
Intelligence,"  /.  Educ.  Psychol.,  1918,  9,  452-456. 

31.  Thomson,  G.  H.   "A  Hierarchy  without  a  General  Factor," 
Brit.  J.  Psychol.,  1916,  8,  271-281. 

32-  •   "A  Worked  Out  Example  of  the  Possible  Linkages  of 


Four  Correlated  Variables  on  the  Sampling  Theory/'  Brit. 
J.  PsychoL,  1927,  1 8,  68-76. 

33.  .    "On  Complete  Families  of  Correlation  Coefficients 

and  Their  Tendency  to  Zero  Tetrad-Differences:  Including 
a  Statement  of  the  Sampling  Theory  of  Abilities,"  Brit.  J. 
PsychoL,  1935,  26,  63-92. 

34.  .    "The  Factorial  Analysis  of  Human  Abilities,"    Hu- 
man Factor,  1935,  9,  180-185. 

35.  Thorndike,  E.  L.,  Educational  Psychology.  N.  Y. :  Teachers 
College,  Columbia  Univ.,  1914.  Vol.  III. 

36.  .  "Intelligence  and  Its  Uses,"  Harper's  Magazine,  1920, 

140,  227-235. 

37.  .     "On  the  Organization  of  Intellect,"  PsychoL  Rev., 

1921,  28,  141-151. 

38.  .    "Intelligence  and  Its  Measurement:  a  Symposium," 

/.  Educ.  PsychoL,  1921,  12,  124-127. 

30^  f     "The  Organization  of  Mind,"  Proceed.,  Forty-third 

Annual  Meeting,  Amer.  PsychoL  Assoc.,  1935,  61-62. 

40.  Thorndike,  E.  L.,  Lay,  W.,  and  Dean,  P.  R.  "The  Relation 
of  Accuracy  in  Sensory  Discrimination  to  General  Intelli- 
gence," Amer.  J.  PsychoL,  1909,  20,  364-369. 

41.  Thorndike,  E.  L.,  et  al.     The  Measurement  of  Intelligence. 
N.  Y.:  Teachers  College,  Columbia  Univ.,  1926.  Pp.  616. 

42.  Thurstone,  L.  L.  Vectors  of  Mind:  Multiple-Factor  Analysis 
for  the  Isolation  of  Primary  Traits.    Chicago:  Univ.  Chicago 
Press,  1935.  Pp.  266. 

43.  Tryon,  R.  C.   "A  Theory  of  Psychological  Components — an 
Alternative   to   'Mathematical   Factors,' "   PsychoL   Rev., 
1935,  42,  42S-454- 



In  Part  I  we  surveyed  some  of  the  major  problems  and 
findings  on  individual  differences  and  attempted  to  un- 
ravel the  factors  and  conditions  which  produce  variation 
from  one  person  to  another.  With  this  background,  we 
may  now  turn  to  an  examination  of  certain  groups  into 
which  individuals  are  commonly  classified.  Such  groupings 
have  been  built  up  through  social  and  cultural  traditions 
and  illustrate  the  general  tendency  to  employ  rigid  cate- 
gories and  sharp  divisions.  Thus  individuals  are  popularly 
classed  into  the  normal  and  the  abnormal,,  the  genius,  the 
feebleminded,  the  insane,  the  neurotic.  Psychological 
differences  are  expected,  or  at  least  sought,  between  the 
sexes  or  among  nations  or  "races."  Many  other  groupings 
can  likewise  be  construed.  A  person  can  be  classified,  for 
example,  in  regard  to  religion,  political  affiliation,  or  even 
place  of  residence.  Psychological  differences  might  be 
expected  between  urban  and  rural  populations,  or  between 
groups  inhabiting  regions  of  different  geographical  char- 
acter, such  as  mountainous  or  flat,  inland  or  coastal,  cold 
or  warm. 

These  various  groupings,  like  all  rigid  classifications  of 
individuals,  are  arbitrary  and  artificial.  In  all  behavioral 
traits,  people  are  distributed  according  to  the  normal  bell- 
shaped  curve  and  cannot  be  assigned  to  distinct  cate- 
gories. When  the  distributions  of  any  two  biologically  or 

O  **.*»««««M».S»»"»«««»^  v**At»    <J»j»U/5i^lrf^4Sa«4f  Jl,n*3».««*l»l»»(4  "HK-VdfSuf'Si-iWi  'wiswS 




averages  of  doubtful  practical  significance.  In  such  com- 
parisons, the  difference  between  the  averages  is  far  smaller 
than  the  range  of  difference  within  either  group.  In  the 
study  of  individuals,  the  only  proper  unit  is  the  individual. 
There  is  no  short-cut  to  the  understanding  of  people,  no 
possibility  of  learning  the  behavioral  peculiarities  of  a  few 
broad  groups  into  which  any  individual  could  then  be 
conveniently  pigeon-holed. 

The  multiple  and  complex  determination  of  the  in- 
dividual's behavioral  development  should  in  itself  make 
us  skeptical  regarding  any  simple  systems  of  characteriz- 
ing people.  Yet  it  is  an  all  too  common  practice  to  expect 
an  individual  to  be  dependable,  or  shiftless,  or  dull,  or 
excitable,  or  poor  in  mechanics,  or  to  ascribe  to  him 
dozens  of  similar  characteristics,  simply  from  the  knowl- 
edge that  such  a  person  is  a  man  or  a  woman,  or  that  he 
belongs  to  a  particular  "race"  or  nation. 

It  is  partly  to  clarify  these  very  muddled  popular  no- 
tions that  the  empirical  study  of  group  differences  ought 
to  be  undertaken.  To  be  sure,  a  careful  examination  of 
the  principles  underlying  individual  variation  in  general 
should  suffice  to  show  the  fallacies  inherent  in  many 
popular  claims  regarding  group  differences.  But  when 
beliefs  are  as  deep-rooted  and  emotionally  tinged  as  those 
governing  many  group  relations,  they  are  not  easily  dis- 
lodged. Direct  evidence  on  the  nature  of  group  differences 
is  more  convincing  than  deductions  from  generally  estab- 
lished principles. 

From  a  more  theoretical  point  of  view,  the  analysis  of 
group  differences  is  a  valuable  adjunct  to  the  investiga- 
tion of  individual  differences  in  general.  The  existence  of 
culturally  diverse  groups  may  be  regarded  as  furnishing  a 
natural  experiment  in  the  production  of  human  variabil- 
ity. If  psychological  differences  among  groups  are  in- 


vestigated  with  reference  to  the  factors  which  brought 
them  about,  the  understanding  of  individual  differences 
will  have  been  considerably  furthered. 

In  the  present  analysis  of  group  differences,  we  shall  be 
more  concerned  with  fundamental  concepts  and  method- 
ological issues  than  with  a  cataloguing  of  results.  The 
latter  have  little  meaning  unless  critically  evaluated.  The 
data  on  group  differences  are  difficult  to  interpret  and  have 
frequently  led  to  opposite  conclusions  in  the  hands  of 
different  writers.  It  is  of  fundamental  importance,  there- 
fore, that  the  special  difficulties  inherent  in  group  com- 
parisons be  clearly  realized  and  that  the  necessary  cautions 
and  controls  be  applied  before  making  any  generaliza- 
tions. With  a  clear  understanding  of  the  problem,  the 
reader  will  be  in  a  position  to  make  his  own  interpretation 
of  any  data  which  he  may  come  across.  And  he  will  also 
be  able  to  guard  against  the  pernicious  habit  of  over- 
hasty  generalization  and  to  detect  the  fallacies  in  errone- 
ous statements  with  which  he  may  be  confronted.  Ob- 
jective and  critical  habits  of  thinking  are  more  urgently 
needed  in  the  field  of  group  differences  than  are  randomly 
gathered  data.  Of  the  latter  there  is  a  sizable  accumula- 
tion, while  the  former  are  all  too  frequently  lacking  even 
in  supposedly  scientific  discussions. 



From  the  earliest  periods  of 

humanlnstory,  we  find  instances  of  conspicuous  deviates 
whom  we  should  now  regard  as  feebleminded  or  insane.1 
Such  individuals  were  generally  considered  to  be  distinct 
beings,  either  representing  a  lower  order  of  humanity  or 

1  For  a  historical  survey  of  conceptions  and  treatment  of  insanity  and  feeble- 
mindedness, see  Farrar  (9)  and  Hollingworth  (14),  Chs.  2  and  3. 


"possessed"  by  spirits.  These  spirits  were  usually  con- 
sidered to  be  evil,  although  in  certain  cases  they  were 
looked  upon  as  gods.  This  demonological  view  was  par- 
ticularly prevalent  during  the  Dark  Ages  and  reflects  the 
influence  of  Christianity  upon  the  thought  of  the  period. 
At  this  time,  it  was  Satan  who  was  thought  to  " possess" 
the  demented  person.  Such  a  condition  was  supposed  to 
be  visited  upon  the  individual  either  as  a  punishment  for 
his  sins  or  as  an  ordeal  to  test  his  moral  fortitude  and 
religious  faith.  Thus  many  persons  who  were  subsequently 
sanctified  by  the  church  experienced  epileptic  seizures, 
hysterical  paralyses  or  anaesthesias,  hallucinations,  para- 
noid delusions,  and  similar  well  known  symptoms  of  in- 
sanity. The  treatment  of  mental  disorders  consisted  of 
exorcism,  physical  abuse,  or  veneration,  depending  upon 
the  particular  theological  doctrine  which  happened  to  be 
current  or  upon  the  specific  circumstances  of  the  subject's 

The  medical  view  of  mental  disorders,  on  the  other  hand, 
was  put  forth  as  early  as  the  fifth  century  B.C.  by  the  Greek 
physician,  Hippocrates.  The  latter  proposed  that  mental 
disorders  result  from  disease  or  injury  to  the  brain.  He 
also  wrote  extensively  on  various  classes  of  mental  dis- 
orders and  their  probable  physical  bases.  For  many  cen- 
turies, the  doctrines  of  Hippocrates  were  accepted  un- 
questionably. They  were  suppressed  during  the  Dark 
Ages,  together  with  most  of  the.  scientific  knowledge  of 
the  Greeks,  but  were  again  resumed  with  the  revival  of 
scientific  interest  and  the  development  of  anatomy  and 
physiology  during  the  Renaissance.  The  medical  con- 
ception of  mental  abnormalities  is  still  prevalent  at  pres- 
ent, especially  among  psychiatrists.1 

The  psychological  study  of  the  abnormal  is  of  relatively 

1 A  psychiatrist  is  a  physician  who  specializes  in  mental  disorders. 


recent  date.  Its  approach  to  the  problem  Is  through  a 
direct  study  of  behavior.  Behavioral  disorders  are  explained 
in  terms  of  behavioral  principles,  rather  than  by  reference 
to  some  other  realm  or  class  of  phenomena.  Abnormal 
psychology  is  an  empirical  and  direct  study  of  extreme 
deviations  in  behavioral  traits.  As  such  it  has  been  re- 
garded as  a  branch  of  differential  psychology. 


The  term  abnormal,  which  means  literally  "away  from 
the  norm/'  has  been  used  to  cover  at  least  three  distinct 
concepts.1  First,  we  may  consider  the  anti-normative  view 
which  regards  as  abnormal  any  deviation  from  the  ideal 
or  perfect  condition.  The  norm  in  such  a  case  is  a  goal  or 
desideratum  to  be  approximated  by  the  existing  conditions. 
To  be  "normal,35  according  to  this  view,  would  be  the 
exception  rather  than  the  rule.  Such  a  use  of  the  terms 
normal  and  abnormal  is  illustrated  by  a  number  of  com- 
mon expressions.  To  say  that  a  person  is  one  of  the 
"lucky  few"  who  have  a  normal  skin  or  normal  teeth,  for 
example,  implies  an  identification  of  normality  with  free- 
dom from  defects  or  other  imperfections.  Or  to  assert 
that  few  can  remain  normal  under  the  stress  and  strain 
of  modern  life  suggests  that  normality  is  an  ideal  state  of 
perfect  composure  and  stability. 

A  second  view  identifies  abnormality  with  pathological 
or  dangerous  conditions.  This  usage  is  particularly  com- 
mon in  medicine.  To  be-  classed  as  abnormal  in  this  sense 
would  have  distinctly  undesirable  connotations.  Such  a 
view  may  be  regarded-  as  an  adaptation  of  the  anti- 
normative  concept  to  meet  practical  and  social  require- 
ments. The  abnormal  still  represents  a  deviation  from  a 
perfect  condition,  but  the  deviation  is  now  so  great  as  to 

1  Cf.  Hollingworth  (14)  and  Foley  (n). 


present  practical  difficulties.  The  condition  requires  action 
of  some  sort  for  the  protection  either  of  the  individual  or 
of  society.  Thus  a  person  who  exhibits  a  few  mild  neurotic 
symptoms,  such  as  a  compulsion  to  avoid  stepping  on  the 
cracks  in  the  sidewalk  or  a  slight  twitching  of  the  forehead, 
may  elicit  the  remark:  "a  bit  queer,  yes — but  not  serious — 
nothing  really  abnormal  about  him."  But  let  such  an  in- 
dividual become  so  depressed  over  an  imagined  or  exag- 
gerated wrong  that  his  work  must  be  discontinued,  or  let 
him  threaten  a  suspected  enemy  with  physical  violence, 
and  he  will  immediately  earn  the  appellation  "  abnormal." 
According  to  this  view,  only  a  very  small  number  of  in- 
dividuals are  abnormal,  the  large  majority  being  indis- 
criminately classified  as  normal. 

Both  of  the  above  views  necessitate  an  arbitrary  norm 
or  standard.  In  the  first,  the  norm  is  a  theoretical  ideal,  in 
the  latter  a  practical  criterion  of  individual  and  social 
survival.  A  more  objective  and  empirical  approach  to,  the 
problem  is  furnished  by  a  purely  statistical, concept  of  ab- 
normality. The  norm  in  this  case  is  the  average.  It  is  the 
usual  and  most  common  condition.  The  abnormal  is  the 
unusual,  the  relatively  infrequent.  The  more  infrequent  a 
condition,  furthermore,  the  more  abnormal  it  is  con- 
sidered. Many  conditions  classed  as  abnormal  in  the 
pathological  sense  would  also  be  regarded  as  statistically 
abnormal  because  of  their  relative  rarity.  On  the  other 
hand,  the  majority  of  those  individuals  classed  as  ab- 
normal according  to  the  anti-normative  view  would  be 
considered  normal,  since  they  constitute  the  large,  inter- 
mediate, and  most  representative  segment  of  the  popula- 
tion. Similarly,  those  who  approximate  the  ideal  or  per- 
fect state  too  closely  would  now  be  regarded  as  abnormal, 
since  they  deviate  significantly  from  the  ordinary,  average 


It  follows  from  the  statistical  view  that  the  abnormal 
may  be  either  inferior  or  superior  to  the  normal.  The 
abnormal  corresponds  simply  to  the  two  ends  of  the  nor- 
mal distribution  curve.  Since  the  curve  is  symmetrical, 
the  superior  deviate  is  just  as  abnormal  as  the  inferior,  in 
the  sense  that  he  is  equally  far  from  the  norm.  It  is 
apparent  that  this  is  the  only  sense  in  which  abnormality 
can  be  objectively  determined  and  measured.  To  speak 
of  inferiority  and  superiority  implies  evaluation  in  terms 
of  specific  biological  and  cultural  requirements.  Such 
evaluation  is  characterized  by  a  certain  degree  of  imper- 
manence  and  subjectivity  which  often  confuse  the  prob- 
lem. A  purely  statistical  concept  of  abnormality,  on  the 
other  hand,  limits  itself  to  an  incontrovertible  and  em- 
pirical criterion. 

It  is  in  the  statistical  sense  that  the  term  abnormal  will 
be  employed  in  the  present  discussion.  In  respect  to  any 
specific  situation,  the  abnormal  will  be  further  subdivided 
into  the  subnormal  and  the  super-normal.  The  present 
chapter  will  be  concerned  with  the  subnormal  deviates. 
The  super-normal  will  be  considered  in  the  following  chap- 
ter on  "genius."  These  two  groups  of  extreme  deviates 
should  be  constantly  viewed  in  their  proper  perspective,  as 
opposite  ends  of  a  continuous  distribution  curve. 

It  is  noteworthy  that  the  terms  abnormal  and  sub- 
normal are  frequently  employed  interchangeably.  In 
everyday  speech,  it  has  become  almost  impossible  to  use 
the  word  abnormal  in  its  innocuous  etymological  sense. 
To  congratulate  a  great  scientist  upon  a  recent  discovery 
by  informing  him  that  we  consider  him  extremely  ab- 
normal would  probably  be  a  breach  of  etiquette.  Nor  is 
this  confusion  restricted  to  popular  usage.  Most  text- 
books on  abnormal  psychology,  for  example,  deal  exclu- 
sively with  the  subnormal,  A  few  make  brief  mention  of 


the  logical  need  for  including  "genius"  in  this  category. 
Having  acknowledged  this  fact,  however,  they  then  devote 
all  subsequent  chapters  to  the  subnormal 

The  identification  of  abnormality  with  subnormality 
may  result  in  part  from  the  influence  of  the  anti-normative 
and  pathological  views.  Such  a  confusion  of  terms  also 
offers  an  interesting  commentary  upon  human  thought.  It 
is  an  all  too  common  practice  to  regard  as  inferior  what- 
ever  differs  from  oneself.  Mutual  racial  prejudices  are  a 
good  example  of  this  tendency.  The  strange  paradox  that 
several  distinct  groups  may  each  regard  themselves  as 
superior  to  all  the  others  is  attributable  to  this  human 
characteristic.  To  be  too  different  from  oneself  must 
surely  imply  some  sort  of  defect,  otherwise  one's  self- 
respect  would  be  too  seriously  impaired! 


Feeblemindedness  represents  the  lower  end  of  the  dis- 
tribution of  intelligence.  It  is  characterized  by  intellectual 
rather  than  emotional  defect.  The  term  feeblemindedness 
is  not  used,  however,  to  cover  deficiency  in  any  ability. 
Thus  an  individual  may  be  far  below  average  in  music, 
drawing,  or  even  mechanical  aptitude,  and  still  be  re- 
garded as  intellectually  normal. 

nateiA  ,  d&jiciency.  *only  ,in  those  abilities, 
essential  /pr  suvvival  .in  our  cultural  ,  milieu  . 

As  was  indicated  in  the  preceding  chapter,  verbal  ability 
probably  plays  the  dominant  role  in  our  conception  of 
feeblemindedness.  Linguistic  aptitude  has,  often  been  ex- 
plicitly accepted  as  a  criterion  of  mental  deficiency.  Thus 
Binet  and  Simon  (4)  wrote:  "An  idiot  is  a  person  who  is 
not  able  to  communicate  with  his  fellows  by  means  of 
language.  He  does  not  talk  at  all  and  does  not  under- 
stand." Similarly,  Esquirol  (cf.  14,  p.  165)  distinguished 


between  three  levels  of  feeblemindedness :  (a)  those  mak- 
ing cries  only;  (b)  those  using  monosyllables;  (c)  those 
using  short  phrases  but  not  elaborate  speech.  Another 
classification  which  is  still  widely  quoted  (cf.  14,  pp.  165- 
166)  is  that  between:  (a)  Idiots,  who  are  incapable  of 
spoken  language,  and  are  limited,  to  gestures;  (b)  Im- 
beciles, who  are  able  to  understand  and  employ  spoken 
language;  and  (c)  Morons,  who  are  capable  of  acquiring 
also  written  language,  but  have  difficulty  with  the  more 
complex  verbal  and  abstract  concepts. 

Feeblemindedness  has  been  described  from  many  points 
oj^vjewrj  Probably  the  most  common  definitions  are  the 
sociological,  or  economic,  and  the  psychometric.  A  widely 
quoted  schema  of  classification  adopted  in  1908  by  the 
Royal  Commission  on  the  Feebleminded,  of  Great  Britain 
(5),  illustrates  the  sociological  conception.  This  classifica- 
tion recognizes  three  grades  of  feeblemindedness,  charac- 
terized as  follows : 

1.  Idiot  (low  grade  amentia) — "A  person  so  deeply  defective 
from  birth  or  from  an  early  age  that  he  is  unable  to  guard 
himself  against  common  physical  dangers." 

2.  Imbecile  (middle  grade  amentia) — "One  who,  by  reason 
of  mental  defect  existing  from  birth  or  from  an  early  age, 
is  incapable  of  earning  his  own  living,  but  is  capable  of 
guarding  himself  against  common  physical  dangers." 

3.  Moron1  (high  grade  amentia) — "One  who  is  capable  of 
earning  a  living  under  favorable  circumstances,  but  who 
is  incapable,  from  mental  defect  existing  from  birth  or 
from  an  early  age,  (a)  of  competing  on  equal  terms  with 
his  normal  fellows,  (b)  of  managing  his  affairs  and  himself 
with  ordinary  prudence." 

1  In  England,  the  term  "feebleminded"  refers  to  the  high  grade  aments,  and 
"amentia"  is  used  as  a  general  term  to  cover  all  degrees  of  mental  deficiency. 
The  term  "moron"  has  been  substituted  for  "feebleminded"  in  the  above  defini- 
tion, in  accordance  with  the  more  familiar  American  usage. 


The  psychometric  classification  is  more  common  among 
mental  testers  and  admits  of  more  quantitative  definition.1 
When  applied  only  to  adults,  the  differentiation  is  often 
made  on  the  basis  of  mental  age.  Thus  an  adult  whose 
mental  age  is  three  years  or  less  is  usually  regarded  as  an 
idiot;  between  three  and  seven  is  the  imbecile  level; 
morons  fall  above  a  mental  age  of  seven  but  fail  to  reach 
the  average  adult  level.  To  make  the  classification  appli- 
cable to  children  as  well  as  adults,  the  limits  have  been 
expressed  in  terms  of  LQ.  Terman's  classification  (23, 
p.  79)  is  probably  the  most  widely  employed  and  has  been 
reproduced  below. 

Designation  LQ. 

Dullness,  rarely  classifiable  as  feeblemindedness  80-90 
Borderline  deficiency,  sometimes  classifiable  as  dull- 
ness, often  as  feeblemindedness  70-80 
Moron  50-70 
Imbecile  20-50 
Idiot  below  20 

It  should  be  borne  in  mind  that  these  distinctions  are 
purely  arbitrary  and  are  made  only  for  practical  con- 
venience. There  is  no  sharp  dividing  line  either  between 
the  normal  and  the  feebleminded,  or  between  the  various 
"levels"  of  feeblemindedness.  The  intellectual  differences 
are  of  degree  only  and  form  a  continuous  gradation,  al- 
though the  social  effects  may  differ  qualitatively. 

The  feebleminded  have  also  been  classified  in  respect  to 
physical  characteristics.  According  to  the  early  medical 
conceptions,  feeblemindedness  was  a  disease  which  was 
expected  to  show  physical  as  well  as  mental  symptoms. 
The  percentage  of  feebleminded  persons  who  present  a 
distinct  morphological  picture  proved,  however,  to  be 
very  small.1  The  large  remaining  majority  had  to  be  given 

1  Hollingworth,  for  example,  estimates  the  number  as  only  10%  (14,  p.  159). 

_  THE  SUBNORMAL  _  337 

the  unrevealing  appellation  of  "  simple  amentia."  The 
major  clinical  varieties  of  feeblemindedness  which  have 
been  differentiated  include  microcephaly,  hydrocephaly, 
cretinism,  and  Mongolism. 

The.^  microcephalic  has  an  abnormally  small,  pointed 
skull,  producing  a  characteristic  "sugar-loaf"  appearance. 
'TEoTiydirocep'halic  has  a  very  large  skull  and  an  excessive 

fluid  intervening  between  the 
-the  brain,  Th&.crftin  has  already  been  described 
in  a  different  connection  (cf.  Ch.  VIII).     He  is  easily 

..  physique,  coarse  thick  skin,  Ipss 

ofjiair,^and  other  physical  characteristics.  This  condition 
has  been  definitely  linked  with  a  thyroid  deficiency  and 
has  been  frequently  relieved  by  repeated  administration  of 
thyroid  extract  in  early  childhood.  Mongolism^  acquired 
•its  name  from  a  certain  facial  resemblance  to  the  Mon- 
golian race.  This  ^similarity,  it  should  be  mentioned,  is 
quite  superficial  and  slight.  The  peculiar,  narrow,  slanting 
"eyes""  cUar  acteris  tic  of  this  type  of  ament  were  chiefly 
responsible  for  the  alleged  resemblance  to  the  Mongolian 
face.  This  condition  can  also  be  distinguished  by  the  shape 
of  the  skull  and  certain  peculiarities  of  the  tongue.1 

Apart  fromjhej;elj^ 
fiable  into  these  special  clinical  varieties,  the 

•~~~*«»**^~~*.  T  "3f  '(a>%-A1w*n1*rP''s'     '  f"m""     "     ' 

are  noT3S¥^^fferentiat;ed  from^mal  i 
traits.^Data  relevant  to  this  question  have  already  been 
presented  in  Chapter  VIII.  It  will  be  recalled  that  in  such 
physical  measures  as  height  and  weight,  the  differences 
between  the  averages  of  normal  and  feebleminded  sub- 
jects were  small  and  the  overlapping  large.  It  is  true  that 
in  practically  every  comparison  the  feebleminded  are  con- 

1  For  a  comprehensive  survey  of  these  clinical  varieties  as  well  as  the  charac- 
teristics of  the  feebleminded  in  general,  the  reader  is  referred  to  Tredgold  (25)  ; 
see  also  22,  for  a  discussion  of  feeblemindedness  in  children. 



sistently  below  normal  in  physical  as  well  as  mental 
traits  (cf.,  e.g.,  8,  20).  But  in  evaluating  these  findings, 
we  must  bear  in  mind  the  influence  of  social  and  economic 
level  upon  both  physical  and  mental  development.  In 
dealing  with  institutional  cases,  furthermore,  the  operation 
of  selective  factors  must  be  taken  into  account.1 

It  has  been  repeatedly  demonstrated  that  the  feeble- 
minded are  not 'equally  deficient  in  all  functions  and  that 
the  degree  of  their  inferiority  increases  as  we  go ,  from 
sinvgle^sensoiy  arid  .motor  traits  to,  complex  intellectual 
processes,  and  especially  "symbolic"  activities  (cf.  6,  14, 
20).  3>Torsworthy?  for  example,  found  that  the  percentage 
of  feebleminded  subjects  out  of  a  group  of  157  who  reached 
or  exceeded  the  median  score  of  normal  groups  was  much 
greater  in  the  simple  sensory  arid  perceptual  functions 
thaji  in  abstract  verbal  tests.  The  same  was  true  when  the 
percentages  who  reached  or  exceeded  the  —  i  P.E.  point 
rather  than  the  median  were  compared.2  The  data  for 
both  types  of  comparison  are  presented  below  (cf.  20,  p.  68). 

Function  Tested 

Comparison  of  weights 
Cancelling  A's 

Cancelling  words  containing  a  and  t 
Memory  for  unrelated  words 
Memory  for  related  words 
Association :  part-whole 
Association:  genus-species 
Association:  opposites,  form  I 
Association:  opposites,  form  2 

Percentage  of  Feebleminded 
Subjects  Who  Reached  or  Exceeded: 

Median  of  —  I  P.E.  of 

Normal  Group  Normal  Group 

18      '  28 

9  18 

I  14 

6  18 

5  19 

9  17 

9  16 

o  0.9 

o  i 

achevments  of 

1  Cf.  Ch.  VIII  for  a  discussion  of  both  points. 

2  In  a  normal  group,  75%  of  the  cases  would  reach  or  exceed  —  I  P.E. 


subjects.  Performance  in  each  subject  was  expressed  as 
a  ratio  of  the  average  achievement  expected  from  normal 
children  of  the  same  age.  These  ratios  are  reproduced 

School  Subject  Educational  Ratio 

Spelling  .460 

Reading  comprehension  .489 

Composition  .514 

Reading:  accuracy  and  speed  .534 

Subtraction  .534 

Division  .553 

Addition  .565 

Multiplication  .589 

Writing:  speed  and  quality  .609 

Drawing  .649 

Handwork  .697 

It  is  apparent  from  this  analysis  of  abilities  along  vari- 

^      >^oup  are  most  de- 

Bartask's.  In  arithmetic,  their  performance  is 
intermediate,  and  in  activities  involving  mechanical  apti- 
tude or  motor  skill,  they  are  closest  to  the  norm.  This 
does  not  mean,  however,  that  such  a  hierarchy  of  defi- 
ciency exists  within  the  individual  feebleminded  person. 
The  same  result  might  follow  if  there  were  more  feeble- 
minded persons  deficient  in  verbal  ability,  fewer  deficient 
in  arithmetic  ability,  and  fewest  in  mechanical  or  motor 
aptitudes.  THe  relationship  between  such  group  averages 
depends  not  only  upon  the  relative  amount  of  inferiority 
displayed  by  each  individual,  but  also  upon  the  number 
of  persons  who  are  inferior.  Since  verbal  aptitude  plays 
such  a  large  part  in  the  criterion  of  feeblemindedness, 
almost  all  persons  in  a  feebleminded  group  will  be  defi- 
nitely below  normal  in  this  trait.  This  consistent  inferior- 
ity will  of  course  produce  a  very  low  group  average  in 
verbal  traits. 


In  arithmetic  aptitude,  many  will  be  below  average, 
some  may  be  normal,  and  a  few  even  superior.  The  slight 
positive  correlation  between  performance  on  arithmetic 
and  verbal  tests,  as  well  as  the  fact  that  arithmetic  tests 
are  frequently  included  in  scales  of  "general  intelligence," 
would  lead  us  to  expect  the  majority,  but  not  all,  feeble- 
minded subjects  to  be  below  the  norms  in  arithmetic 
aptitude.  This  would  result  in  a  group  average  higher 
than  that  in  verbal  traits,  but  still  considerably  below 

In  tasks  involving  motor  skills  or  aptitude  in  mechanics, 
music,  or  pictorial  art,  we  should  expect  the  feebleminded 
to  be  distributed  very  similarly  to  a  normal  group,  since 
these  traits  show  little,  or  no  correlation  with  verbal  ability 
or  intelligence  test  performance.  The  majority  would  in 
this  case  be  normal,  only  a  small  number  inferior,  and  a 
few  superior.  As  a  result,  the  status  of  the  group  as  a 
whole  would  be  only  slightly  below  normal.  Thus  it  is 
apparent ,  that  the,hlerarchy  of  deficiency  usually  found  la 
feebleminded  groups  is  #  result  of  the  culturally  imposed 
criterion  of  feeblemindedness  as  well  as  the  organization 
of,  abilities. 


Insanity  consists  of  a  marked  deviation  in  emotional 
or  other  personality  traits.  The  individual,  although  in- 
tellectually normal  or  even  superior,  is  unable  to  make  a 
satisfactory  adjustment  because  of  serious  personality 
disorders.  Thus  he  may  have  delusions  of  persecution, 
which  make  him  suspect  all  with  whom  he  comes  into 
contact  of  plotting  against  him,  or  delusions  of  grandeur, 
in  which  he  believes  himself  to  be  Napoleon  or  some  other 
favorite  character.  Such  symptoms  are  classified  under 
paranoia.  Or  he  may  develop  such  extreme  introverted 


tendencies  as  to  lose  all  contact  with  his  fellow-beings  and 
with  occurrences  about  him,  as  in  dementia  prtecox.  His 
symptoms  may  consist  of  recurrent  periods  of  extreme 
depression  and  excitement,  as  in  manic-depressive  insan- 
ity. These  examples  will  suffice  to  show  the  varied  forms 
which  insanity  may  take.  We  are  not  here  concerned  with 
a  classification  of  insanities,  or  psychoses,  into  the  common 
clinical  varieties,  nor  with  an  enumeration  of  symptoms. 
This  material  can  be  found  in  any  standard  textbook  of 
psychiatry  and  in  many  books  on  abnormal  psychology 
(cf.,  e.g.,  17,  i85  19).  We  are  concerned  only  with  the  gen- 
eral psychological  nature  of  such  psychoses. 

In  the  first  place,  insanity  is  not  to  be  confused  with 
feeblemindedness.  The  insane  are  recruited  from  all  in- 
tellectual levels.  Many  psychotics  fall  within  the  normal 
range  of  intelligence.  Instances  are  not  unknown  among 
the  highly  gifted.  And  likewise,  insanity  may  occur  in 
feebleminded  individuals,  although  for  some  psychoses, 
such  as  paranoia,  a  certain  degree  of  complexity  of  men- 
tal life  seems  to  be  required.  A  number  of  psychotic  con- 
ditions, such  as  dementia  prsecox,  lead  to  intellectual 
deterioration,  but  there  are  others,  such  as  paranoia,  in 
which  the  subject  may  suffer  no  impairment  of  abilities. 

As  in  the  case  of  feeblemindedness,  there  is  no  sharp 
dividing  line  between  insanity  and  normality.  Sharp  dis- 
tinctions are  made  for  practical  purposes  of  confinement, 
treatment,  and  similar  reasons,  but  close  examination 
reveals  a  continuous,  unbroken  gradation  from  the  thor- 
oughly well-adjusted  person  to  the  conspicuously  insane. 
Psychotic  symptoms  differ  only  in  degree  from  the  behav- 
ioral peculiarities  of  the  normal  individual.  From  the 
blissful  optimist  who  trusts  implicitly  whomever  he  meets 
to  the  paranoiac  who  believes  that  the  stranger  who  acci- 
dentally brushes  against  him  is  plotting  his  demise,  there 


are  all  degrees  of  "suspiciousness."  The  same  may  be  said 
of  all  other  characteristics  of  the  insane.  A  good  example 
of  this  is  the  familiar  case  of  the  student  who,  upon  read- 
ing a  manual  of  psychiatry  or  attending  a  course  in  ab- 
normal psychology,  believes  himself  to  be  afflicted  with 
each  form  of  psychosis  in  turn.  Most  of  us  can  discover 
in  ourselves  at  least  one  characteristic  of  many  types  of 
insanity,  in  mild  form.  It  is  not  normal,  in  the  statistical 
sense,  to  be  entirely  free  from  all  such  slight  peculiarities. 

A  fundamental  distinction  from  the  viewpoint  of  the 
psychologist  is  that  between  organic  and  functional  dis- 
orders. Briefly,  organic  disturbances  are  those  which  can 
be  definitely  attributed  to  a  structural  deficiency.  In 
functional  disorders,  on  the  other  hand,  there  seems  to  be 
only  a  faulty  operation  or  deficient  action  of  apparently 
normal  structures.1  Thus  general  paresis  has  been  defi- 
nitely traced  to  the  influence  of  syphilitic  infection  upon 
the  nervous  system;  a  group  of  psychoses  have  been  shown 
to  develop  from  excessive  use  of  alcohol  or  drugs;  injuries, 
or  lesions,  to  certain  parts  of  the  brain  or  lower  nerve 
centers  lead  to  characteristic  behavioral  deficiencies.  There 
remain,  however,  a  large  number  of  psychotic  conditions 
for  which  no  physical  basis  has  been  discovered.  Some 
claim  that  the  physical  concomitants  of  such  behavioral 
disorders  are  only  undiscovered  and  that  ultimately  all 
will  be  adequately  explained  in  structural  terms.2 

There   is    a   growing   conviction,    however,    especially 

1  This  distinction,  although  generally  employed  only  with  reference  to  per- 
sonality disorders,  could  be  applied  equally  well  to  feeblemindedness.  Cretinism, 
for  example,  represents  an  organic  mental  deficiency  resulting  from  physical  dis- 
orders. In  nearly  90%  of  feebleminded  cases,  however,  no  structural  deficiency 
can  be  discovered  and  the  disturbance  might  be  regarded  as  functional.   In  a  few 
instances,  as  in  isolation  amentia,  the  functional  nature  of  the  disorder  is  clearly 

2  For  a  discussion  of  various  forms  of  insanity  from  this  point  of  view,  see 
Moss  and  Hunt  (19). 


among  psychologists,  that  these  disorders  may  be  purely 
functional  in  their  origin  and  involve  no  structural  im- 
pairment. If  this  be  the  case,  we  should  seek  the  causes 
of  psychotic  conditions  in  the  mechanism  of  learning  and 
in  the  environmental  conditions  which  have  surrounded 
the  individual.1  This  question  is  still  a  highly  controver- 
sial one  to  which  no  final  answer  can  be  given  as  yet. 

That  many  behavioral  disorders  are  functional  can, 
however,  be  clearly  demonstrated.  A  number  of  tests 
have  been  suggested  for  diagnosing  the  functional  nature 
of  several  types  of  symptoms  (cf.  14,  Ch.  10).  Thus  func- 
tional disorders  are  more  often  intermittent  and  occasional 
than  organic  disorders  and  may  depend  upon  the  presence 
of  a  specific  situation  or  stimulus.  The  symptom  may  only 
be  manifested,  for  example,  in  the  presence  of  certain 
individuals  or  in  a  particular  locality.  Functional  dis- 
orders are  also  more  variable,  exhibiting  more  individual 
idiosyncracy  from  one  case  to  another.  Functional  symp- 
toms of  a  motor  or  sensory  nature  can  usually  be  detected 
because  of  their  anatomical  impossibility.  The  so-called 
"stocking  and  glove"  anaesthesia,  for  example,  is  a  loss  of 
sensitivity  in  regions  which  are  unitary  only  in  popular 
thought,  but  do  not  correspond  to  any  neurologically  pos- 
sible division.  In  certain  types  of  symptoms,  special 
methods  of  examination,  employing  unusual  situations 
unknown  to  the  subject,  will  also  reveal  the  functional 
nature  of  the  disorder. 

Finally,  a  very  striking  characteristic  of  functional  dis- 
orders is  their  susceptibility  to  various  fantastic  methods 
of  treatment.  In  general,  nearly  any  conceivable  procedure 
or  technique  may  cure  such  symptoms,  provided  the  patient 
is  convinced  of  its  efficacy.  The  success  of  many  systems  of 

1  For  a  vivid  analysis  of  the  method  whereby  such  disorders  might  be  built  up 
through  learning,  cf.  Watson  (26). 


mental  therapy  which  have  achieved  widespread  publicity 
can  be  attributed  to  this  peculiar  characteristic  of  func- 
tional symptoms.  Similarly,  the  sudden,  startling,  and 
miraculous  " cures"  effected  at  various  shrines,  one  of  the 
best  known  of  which  is  that  of  Sainte  Anne  de  Beaupre  in 
Canada,  testify  to  the  functional  nature  of  the  disorders. 
Dramatic  incidents  are  recounted  of  the  blind  who  sud- 
denly regained  their  vision,  or  of  the  crippled  who  aban- 
doned their  crutches  and  walked  forth  erect.  The  same 
results  have  occasionally  been  observed  in  a  severe  emer- 
gency, as  when  a  supposedly  hopeless  cripple  arises  from 
his  chair  to  save  a  loved  one  from  drowning  or  fire. 

The  history  of  such  cases  usually  reveals  a  certain  ob- 
scurity of  diagnosis.  Physicians  were  baffled,  the  case  was 
declared  hopeless  or  a  mystery,  thus  adding  to  the  glamor 
of  the  "cure."  What  this  suggests,  however,  is  the  func- 
tional nature  of  the  disorder  which  naturally  defied  or- 
ganic diagnosis  and  treatment. 

In  closing  this  brief  survey,  mention  should  be  made  of 
the  fact  that  these  disorders  are  no  less  "real"  because 
they  are  functional.  The  subject  may  suffer  just  as  acutely 
and  be  as  seriously  handicapped  as  if  he  had  a  definite 
structural  deficiency.  Similarly,  such  disturbances  can- 
not be  overcome  merely  by  voluntary  effort.  Nor  should 
they  be  confused  with  malingering.  The  subject  himself 
is  undergoing  as  vivid  an  experience  as  if  he  had  an  organic 
disorder  and  he  may  be  completely  unaware  of  the  fact 
that  his  symptoms  have  no  structural  basis. 


Neuroses  may  be  regarded  as  milder  forms  of  personal- 
ity disorder  than  the  psychoses.  They  are  also  more  gen- 
erally regarded  as  functional  in  their  nature  than  the 
latter.  In  their  specific  manifestations,  they  bridge  the 



gap  between  the  slightly  maladjusted  and  emotionally 
unstable  individual  on  the  one  hand  and  the  distinctly 
insane  on  the  other.  On  the  basis  of  the  normal  distribu- 
tion of  behavioral  traits,  we  should  expect  neurotics  to  be 
more  numerous  than  psychotics,  since  they  are  nearer  the 
center  of  the  curve.  This  seems  to  be  borne  out  by  obser- 
vation, although  it  is  difficult  to  obtain  accurate  data  on 
the  relative  incidence  of  such  disorders  because  of  loose 
methods  of  diagnosis. 

A  common  classification  of  neuroses  recognizes  three 
major  types  differentiated  on  the  basis  of  symptomatology, 
wz.,  psychasthenia,  neurasthenia,  and  hysteria.  Typical 
psychasthenic  symptoms  are  phobias,  or  abnormal  fears, 
compulsions,  obsessions,  worries,  restlessness,  and  irrita- 
bility. Neurasthenia  is  characterized  by  such  symptoms  as 
dizzy  spells,  headaches,  excessive  feelings  of  fatigue,  di- 
gestive disturbances,  and  the  like.  Hysteria  covers  motor 
and  sensory  disturbances,  such  as  tremors,  contractions, 
paralyses,  loss  of  sensation,  and  heightened  sensitivity,  as 
as  well  as  fits  of  unconsciousness,  loss  of  memory,  and 
similar  somnambulistic  disorders.  It  has  been  suggested 
(cf.  14,  pp.  403-404)  that  psychasthenia  represents  a  dis- 
order on  the  intellectual,  or  symbolic,  level,  neurasthenia 
on  the  autonomic,  or  visceral,  and  hysteria  on  the  postural. 

Neuroses,  like  psychoses,  bear  little  relation  to  intellec- 
tual level.  Neurotic  symptoms  occur  over  a  wide  range  of 
mentality.  In  Table  XV  will  be  found  the  mental  ages  of 
.1172  soldiers  in  the  United  States  army  during  the  World 
War  who  manifested  various  neurotic  disorders*  These 
mental  ages  were  determined  by  a  variety  of  ways,  in- 
cluding Stanford-Binet,  performance  scales,  Army  Alpha, 
a  series  of  five  specially  selected  tests,1  or  a  combination 
of  more  than  one  of  these  methods. 

1  Completion,  opposites,  substitution,  word-building,  and  digit  span. 




(After  Hollingworth,  13,  p.  87) 








nosed  \ 

















































































































Over  1  8 
















*  Not  classifiable  into  a  specific  sub-form. 

f  Undiagnosed  at  the  time  of  testing,  but  believed  to  be  chiefly  neurotic. 

It  will  be  noted  that  all  of  the  disorders  occur  over  the 
entire  range  of  intelligence  test  scores,  except  psychas- 
thenia,  which  seems  to  be  absent  below  a  mental  age  of 
ten  years.  Thej^rious  neurotic  groups  tend,  however,  to 

*f5  scale.    It  has  been  sug- 

gested  on  t:his.,tesis  that  intellectual  ^ 

certain  extent  thep|^^  of  neurotic  symptom  which  an 
individual'  will  develop.  Thus  duller  subjects  are  expected 
to  "Be  more  susceptible  to  disorders  of  the  hysteric  type 
and  brighter  subjects  to  psychasthenia;  neurasthenia 
occupies  an  intermediate  position.  The  army  data  given 


in  Table  XV  yield  the  following  median  mental  ages  for 
these  three  groups. 

Diagnosis  Nu™bes  °/  ****** 

Cases  M.A. 

Hysteria  177  11.5 

Neurasthenia  83  13.0 

Psychasthenia 1  10  14.0 

In  evaluating  these  group  tendencies,  one  must  not  lose 
sight  of  the  fact  that  there  is  much  overlapping  between 
the  groups  and  wide  variation  of  intellectual  level  within 
each  type  of  neurosis.  The  group  differences  themselves 
may  be  variously  interpreted.  Environmental  factors, 
for  example,  might  prove  adequate  to  account  for  the 
slight  relationship  found  between  intelligence  and  nature 
of  neurotic  symptom.  Individuals  of  lower  intellectual 
status  usually  come  from  inferior  social,  economic,  and 
educational  levels.  The  problems  and  difficulties  which 
they  encounter  are  thus  of  a  different  nature  from  those 
which  confront  individuals  in  a  superior  environment. 
The  emotional  disorders  developed  by  such  groups  can 
reasonably  be  expected  to  differ  accordingly. 

One  further  point  deserves  mention  in  connection  with 
the  neuroses.  One  often  hears  the  statement  that  neuroses 
are  a  product  of  the  strain  and  stress  of  civilized  life. 
The  evidence  cited  is  the  appalling  number  of  persons, 
now  classed  as  neurotics,  who  visit  psychiatrists  or  spend 
brief  periods  "recuperating"  in  sanatoriums.  History, 
it  is  pointed  out,  bears  little  testimony  to  the  presence 
of  this  class  of  individuals  in  former  eras.  The  fallacy 
inherent  in  this  argument  is  a  common  one.  With  the 
rapid  development  of  psychology  and  psychiatry,  there 
has  come  a  more  general  recognition  of  neurotic  conditions. 

1  It  is  of  course  difficult  to  generalize  from  only  10  cases,  but  data  of  other 
investigators  agree  in  showing  psychasthenics  to  have  higher  intelligence  test 
scores  than  other  neurotics. 


These  mild  abnormalities  may  have  always  existed,  but 
they  passed  undiagnosed  until  the  present.  The  neurotic 
is  not  sufficiently  maladjusted  in  most  cases  to  attract 
much  attention  or  demand  urgent  treatment.  In  the 
past,  such  individuals  may  have  gone  their  unhappy  way, 
probably  unpopular  or  disliked  among  their  fellow-beings, 
but  bearing  their  difficulties  unlabeled  and  unrecorded. 
Similarly,  the  more  highly  developed  the  methods  of 
diagnosis,  the  milder  will  be  the  disorders  which  are 
detected  and  the  more  numerous  the  individuals  classified 
as  neurotic. 


Recent  interest  in  the  study  of  abnormal  conditions  in 
widely  varying  cultural  milieux  hasjrevealedjJ^^ 
abnormality;  is  culturally  defined.1  There  are  two  distinct 
ways  in  which  specific  cultural  standards  determine  what 
shall  be  considered  abnormal.  First,  the  norm  will  differ 
from  one  group  to  another,  so  that  any  one  specific  be- 
havioral manifestation  may  occupy  entirely  different  posi- 
tions in  different  distribution  curves.  An  example  from 
physical  traits  will  make  this  clear.  If  we  ask  whether  a 
given  individual  is  "tall"  or  "short,"  we  may  obtain 
very  different  answers  when  different  groups  are  employed 
as  a  standard.  The  same  individual  might  be  abnormally 
tall  when  evaluated  in  reference  to  the  distribution  of 
Japanese  subjects,  and  abnormally  short  in  terms  of 
the  distribution  of  Scandinavians.  Similarly,  in  certain 
groups  violent  displays  of  emotions  are  the  rule  and 
stolidity  would  be  abnormal.  In  others,  the  reverse  is 

Secondly,. .  cultui#]Jy,.Mest^ 
which  end  off  the,  .distribution  -cwr ^•^fAd^'^uto^z^^J"^^/'-- 

1  For  a  comprehensive  and  systematic  analysis  of  this  problem,  cf.  Foley  (il). 


normal  and  which  end  subnormal.  As  has  already  been 
pointed  out,  such  a  criterion  is  intrinsically  cultural  in 
its  nature.  Comparative  anthropology  furnishes  many 
instances  of  behavioral  deviations  which  are  regarded  as 
unadaptive,  pathological,  insane,  or  mentally  deficient 
in  one  culture  and  are  admired  or  revered  in  another. 
Such  behavior  may  be  abnormal  in  both  cases,  in  the 
statistical  sense,  but  its  social  evaluation  and  practical 
value  in  the  different  cultures  place  it  at  opposite  ends 
of  the  scale.  This  point  has  been  clearly  expressed  by 
Benedict  (i)  who  states: 

...  it  is  probable  that  about  the  same  range  of  individual 
temperaments  are  found  in  any  group,  but  the  group  has 
already  made  its  cultural  choice  of  those  human  endowments 
and  peculiarities  it  will  put  to  use  .  .  .  the  misfit  is  the 
person  whose  disposition  is  not  capitalized  by  his  culture.  .  .  . 
It  is  clear  that  there  is  not  possible  any  generalized  descrip- 
tion of  "the"  deviant — he  is  the  representative  of  that  arc  of 
human  capacities  that  is  not  capitalized  in  his  culture  (p.  24). 

The  same  point  of  view  is  further  elaborated  in  a  later 
article  (2)  by  Benedict,  in  which  appear  the  following 
statements : 

One  of  these  problems  relates  to  the  customary  normal- 
abnormal  categories  and  our  conclusions  regarding  them.  In 
how  far  are  such  categories  culturally  determined,  or  in  how 
far  can  we  with  assurance  regard  them  as  absolute?  In  how 
far  can  we  regard  inability  to  function  socially  as  diagnostic 
of  abnormality,  or  in  how  far  is  it  necessary  to  regard  this 
as  a  function  of  the  culture? 

As  a  matter  of  fact,  one  of  the  most  striking  facts  that 
emerge  from  a  study  of  widely  varying  cultures  is  the  ease 
with  which  our  abnormals  function  in  other  cultures.  It 
does  not  matter  what  kind  of  "abnormality"  we  choose  for 
illustration,  those  which  indicate  extreme  instability,  or  those 


which  are  more  in  the  nature  of  character  traits  like  sadism 
or  delusions  of  grandeur  or  of  persecution,  there  are  well 
described  cultures  in  which  these  abnormals  function  at  ease 
and  with  honor,  and  apparently  without  danger  or  difficulty 
to  the  society  (p.  60). 

As  an  example  of  behavior  regarded  as  subnormal  in 
our  society,  but  valued  and  respected  in  other  cultures, 
may  be  mentioned  the  conventional  behavior  of  the  Si- 
berian shaman,  or  priest  (cf.  7).  Such  phenomena  as 
cataleptic  seizures  often  constitute  an  essential  component 
of  the  behavior  of  the  shaman.  Similarly,  trances,  ex- 
treme introversion  bordering  on  dementia  praecox,  manic- 
depressive  behavior,  and  homosexuality  are  fostered  and 
considered  the  ideal  norm  of  behavior  in  various  cul- 
tures. Paranoia,  with  its  delusions  of  grandeur  and  its 
extreme  suspiciousness,  may  be  identified  in  the  behavior 
and  traditional  chants  of  many  medicine  men  or  tribal 
priests.  Illustrations  could  easily  be  multiplied  from  a 
survey  of  the  behavior  of  different  peoples.1  - 


It  should  also  be  noted  that  behavioral  deviations  suf- 
ficiently pronounced  to  be  classified  as  abnormal  in  the 
statistical  sense  are  not  an  exclusively  human  phenome- 
non. Marked  deviates  can  be  found  in  all  species.  Mental 
deficiency,  as  well  as  "unadaptive"  behavior  which  may 
be  characterized  as  psychotic  or  neurotic,  has  been  ob- 
served in  many  animal  forms.  Abnormal  reactions  rang- 
ing from  drowsiness  and  sleep  to  overexcitability  and 
violent  emotional  displays  have  been  observed  in  chim- 
panzees when  the  latter  were  confronted  with  an  "  over- 
exacting"  situation  (cf.  3,  Ch.  VIII).  Similarly,  Pavlov 
(21)  observed  distinctly  neurotic  behavior  in  dogs  in  the 

1  Cf.  Benedict  (i,  2)  and  Foley  (n). 

_  _  THE  SUBNORMAL  351 

course  of  conditioning  experiments.  Symptoms  not  un- 
like those  of  the  human  "nervous  breakdown"  appeared 
when  the  animal  was  required  to  make  too  fine  a  sensory 
discrimination,  or  to  set  up  too  many  conditioned  reac- 
tions within  a  short  time,  or  to  establish  a  conditioned 
reaction  when  the  two  stimuli  were  separated  by  too  long 
an  interval. 

Homosexuality  has  been  observed  or  experimentally 
induced  among  doves,  pigeons,  guinea  pigs,  white  rats, 
and  monkeys  (cf.  12,  15).  Several  investigators  (cf., 
e.g.,  10,  12,  24)  have  reported  instances  of  other  types  of 
abnormal  behavior  among  monkeys,  such  as  habit  residu- 
als, temper  tantrums,  infantile  reversions,  and  various 
forms  of  sexual  perversions.  These  constitute  abnormal- 
ities in  the  sense  that  they  differ  conspicuously  from  the 
usual  behavior  of  the  given  species.  Whenever  the  etiology 
of  such  abnormal  behavior  could  be  definitely  traced, 
experiential  or  environmental  factors  were  again  found 
to  play  a  predominant  part  (cf.  10,  12,  15,  24). 


We  may  now  collate  the  diverse  facts  of  abnormal  be- 
havior reported  in  the  preceding  sections  and  attempt  to 
characterize  the  nature  of  abnormality.  First^  jJiere  is 
nothing  absolute  about  abnormality.  The  deviations 

,  ^.jK.a^w»^'«^f«''v^^ 

termed  '"  abnormal  are  relative  and  continuous.  There  is 
no  sharp  or  clear  dividing  line  between  the  normal  and 
the  abnormal,  nor  do  abnormal  symptoms  differ  qualita- 
tively from  normal  behavior.  Such  symptoms  uare  .only 

exaggeratiqjais,  pi^Q^jmli^b^iorr'*^*"'  *"**" 

Secondly,  abnormality  is  specific  (cf.  n).  Thejndj- 
viduaf  may'T^  trait  and  yet  Remain 

^^ft.^JL***.  ±u**&M~+ja#ilt  &#»>.  Vt!llJ  •„.  -  ,    ,'  „,,.„  iMtftil^'iW*wM'<Sfc-i  .&V*tAt*rt?#«*2*»*'** 

u*M~+ja#ilt  &#»>.  Vt!llJ  •„.  -  ,    ,'  „,,.„  iMtftil^'iW*w'<-  .* 

Thls  1S  true  of 
both  intellectual  and  emotional  traits  and  follows  from 


the  organization  of  behavior  traits  within  the  individual 
(cf.  Chs.  X  and  XI). 

Thirdly,  psychological  abnormality  is  to  be  defined  in 
terms  of  Bebayior.  In  some  cases,  behavioral  disorders 
may  have  structural  concomitants,  such  as  physical  dis- 
eases, lesions,  and  malformations.  But  in  the  majority 
of  cases  no  such  physical  basis  can  be  discovered  and  it 
would  only  obscure  the  issue  to  attribute  such  behavioral 
manifestations  to  unknown  organic  causes.  Analysis  of 
the  behavioral  history  and  environmental  stimulation  of 
the  individual,  on  the  other  hand,  often  reveals  an  ade- 
quate background  for  the  development  of  the  particular 
symptoms.  Functional  abnormalities  are  the  special  do- 
main of  the  psychologist  and  should  be  studied  directly, 
without  vague,  hypothetical  reference  to  physical  con- 

Fourthly,  abnormal  behavior  involves  the  same  psy- 
chological principles  as  normal  behavior.  Abnormality 
i$  the  normal  result  of  certain  stimulatin^Qflditions .  Such 
behavior  is  termed  abnormal  only  because  it  conflicts 
with  the  needs  or  standards  imposed  by  a  given  culture. 
This  is  illustrated  by  two  facts  of  cultural  anthropology, 
viz.,  the  shifting  norm  and  the  cultural  evaluation  of  the 
direction  of  deviation. 


1.  Benedict,  R.    "Configurations  of  Culture  in  North  Amer- 
ica," Amer.  Anthr.,  1932,  N.S.  34,  1-27. 

2.  .   "Anthropology  and  the  Abnormal,"  /.  Gen.  PsychoL, 

1934,  10,  59-82. 

3.  Bentley,  I.  M.     The  Problem  of  Mental  Disorder.    N.  Y.: 
McGraw-Hill,  1934.  Pp.  388. 

4.  Binet,  A.,  and  Simon,  Th.     The  Intelligence  of  the  Feeble- 
minded (transl.  by  Kite,  E.  S.).    Vineland,  N.  J.:  Training 
School  Publ.  No.  n,  1916.  Pp.336. 


5.  "British  Royal  Commission  on  the  Care  and  Control  of  the 
Feebleminded,"  Report  of  the  Commission.    London:  Wy- 
man,  1908. 

6.  Burt,  C.  Mental  and  Scholastic  Tests.  London:  King,  1921. 

7.  Czaplicka,  M.  A.     Aboriginal  Siberia:  a  Study  in  Social 
Anthropology.  Oxford:  Clarendon  Press,  1914.  Pp.334. 

8.  Doll,  E.  A.    Anthropometry  as  an  Aid  to  Mental  Diagnosis. 
Vineland,  N.  J.:  Training  School  Publ.  No.  8,  1916.  Pp.  91. 

9.  Farrar,  C.  B.     "Some  Origins  in  Psychiatry,"  Amer.  J. 
Insanity,  1907-08,  64,  523-552;  1908-09,  65,  83-101;  1909- 
10,  66,  277-294. 

10.  Foley,  J.  P.,  Jr.    "Second  Year  Development  of  a  Rhesus 
Monkey  (Macaca  mulatto)  Reared  in  Isolation  during  the 
First  Eighteen  Months,"  /.  Genet.  PsychoL,  1935,  47,  73-97. 

11.  .    "The  Criterion  of  Abnormality,"  /.  Abn.  and  Soc. 

Psychol,  1935,  30,  279-291. 

12.  Hamilton,  G.  V.   An  Introduction  to  Objective  Psychopathol- 
ogy.  St.  Louis:  Mosby,  1925.  Pp.  354. 

13.  Hollingworth,  H.  L.   The  Psychology  of  Functional  Neuroses. 
N.  Y.:  Appleton,  1920.  Pp.  259. 

14.  .     Abnormal  Psychology:  Its  Concepts  and  Theories. 

N.  Y.:  Ronald  Press,  1930.   Pp.  590. 

15.  Jenkins,  M.  "The  Effect  of  Segregation  on  the  Sex  Behavior 
of  the  White  Rat  as  Measured  by  the  Obstruction  Method," 
Genet.  PsychoL  Mon.,  1928,  3,  457-571. 

1 6.  Klineberg,  0.    Race  Differences.   N.  Y.:  Harper,  1935.    Pp. 


17.  Kraepelin,  E,    Clinical  Psychiatry:  a  Textbook  for  Students 
and  Physicians  (abstr.  and  adapt,  from  Ger.  ed.  by  A.  R. 
Diefendorf).  N.  Y.:  Macmillan,  1918.  Pp.562. 

1 8.  Morgan,  J.  J.  B.      The  Psychology  of  Abnormal  People. 
N.  Y.:  Longmans,  Green,  1932.  Pp.  627. 

19.  Moss,  F.  A,,  and  Hunt,  T.    Foundations  of  Abnormal  Psy- 
chology. N.  Y.:  Prentice-Hall,  1932.   Pp.  548. 

20.  Norsworthy,  N.    "The  Psychology  of  Mentally  Deficient 
Children,"  Arch.  PsychoL,  No.  I,  1906.  Pp.  ill. 


21.  Pavlov,   I.      Conditioned  Reflexes:  an  Investigation   of  the 
Physiological  Activity  of  the  Cerebral  Cortex  (transl.  and  ed. 
by  G.  V.  Anrep).    Oxford  Univ,  Press:  Humphrey  Milford, 
1927.  Pp.  430. 

22.  Pintner,  R.  "  The  Feebleminded  Child."  Ch.  20  in  Handbook 
of  Child  Psychology.    C.  Murchison,  ed.    Worcester,  Mass.: 
Clark  Univ.  Press,  1933.   Pp.  956. 

23.  Terman,  L.  M.     The  Measurement  of  Intelligence.    N.  Y.: 
Houghton  Mifflin,  1916.   Pp.  362. 

24.  Tinklepaugh,  O.  L.    "The  Self-Mutilation  of  a  Male  Ma- 
cacus  rhesus  Monkey,"  /.  Mammal.  ^  1928,  9,  293-300. 

25.  Tredgold,  A.  F.    Mental  Deficiency.    N.  Y.:  Wood,  1922. 
Pp.  569- 

26.  Watson,  J.  B.    "Behavior  and  the  Concept  of  Mental  Dis- 
sease,"  /.  Phil.,  PsychoL,  and  Sci.  Method,  1916,  13, 


The  existence  of  genius  has  probably  been  recognized 
by  man  from  earliest  times.  In  order  to  be  popularly 
acclaimed  as  a  genius,  however,  an  individual  must  possess 
very  exceptional  talents  of  the  kind  demanded  by  his 
culture.  Since  only  the  extreme  deviates  attract  notice, 
they  seem  by  the  very  rare  quality  of  their  attainments 
to  stand  off  from  the  rest  of  mankind  and  constitute  a 
distinct  group.  With  the  advent  of  more  objective  meth- 
ods of  observation  and  the  development  of  testing 
techniques,  the  presence  of  lesser  deviates  who  bridge 
the  gap  between  the  average  man  and  the  person  of  rare 
gifts  has  been  demonstrated.  Thus  the  popular  belief  in 
genius  as  a  separate  species  arose  in  the  same  fashion  as 
the  similar  belief  regarding  the  feebleminded  and  it  is 
being  dispelled  by  the  same  methods. 

The  relationship  between  genius  and  eminence  is  a 
curious  one.  Many  writers  identify  the  two  by  the  simple 
expedient  of  defining  genius  as  the  possession  of  "what  it 
takes"  to  become  eminent  in  our  society.  The  eminent 
man  is  then  considered  a  genius  ipso  facto*  There  would 
thus  be  as  many-kinds  of  genius  as  there  are  ways  of -suc- 
ceeding in  the  particular  society.  The  successful  financier, 
for  example,  may  be  awarded  an  honorary  university 
degree  for  his- "financial  genius,"  the  victorious-  general 
for  his  "military  genius."-  Society  often- creates  a  -new 
form  of  "genius"  in  order  to  rationalize  its  allotment  of 

Almost  any  theory  regarding  the  nature  of  genius  could, 



of  course,  be  defended  by  restricting  the  term  genius  in 
some  arbitrary  way.  The  broadest  and  most  objective 
definition  of  genius  is  that  of  an  individual  who  excels 
markedly  the  average  performance  in  any  field.  Social 
evaluation,  however,  invariably  enters  into  the  concept. 
Genius  is  defined  in  terms  of  specific  social  criteria  and  a 
cultural  sense  of  values.  In  our  society  the  more  abstract 
and  linguistic  abilities  are  considered  the  "higher"  mental 
processes.  Similarly,  certain  lines  of  achievement  enable 
the  individual  to  earn  the  appellation  of  genius  much 
more  readily  than  others.  Thus  academic  and  scientific 
work,  literature,  music,  and  the  plastic  arts  are  rated 
higher  than,  let  us  say,  roller  skating  or  cooking. 

To  be  sure,  very  exceptional  accomplishments  in  the 
latter  fields  might  be  recognized  as  genius,  after  a  fashion. 
An  internationally  famed  roller  skate  acrobat  or  a  re- 
nowned chef-de-cuisine  might  be  called  a  genius  and 
ranked  higher  than  a  mediocre  scientist  or  painter.  But 
in  the  former  instances,  the  attainments  must  be  propor- 
tionately far  greater  than  in  the  latter  in  order  that  the 
individual  may  be  designated  a  genius.  And  even  when 
the  term  genius  is  applied  to  such  cases,  one  feels  that  it  is 
done  only  by  courtesy  and  that  the  word  is  implicitly 
enclosed  in  quotation  marks.  It  is  apparent,  therefore, 
that  In  order  to  have  practical  meaning,  any  definition  of 
genius  must  recognize  the  selection  of  significant  talents 
which  has  been  made  within  a  given  cultural  group. 

A  further  question  which  has  been  vigorously  debated 
is  that_olj^faLK^^  Is  the  man  of 

genius  one  who  manifests  a  well-rounded  intellectual 
superiority  or  one  who  possesses  a  highly  specialized  gift? 
It  follows  from  the  organization  of  mental  traits  that  this 
distinction  is  not  a  valid  one.  Siqcs  th&  iuteicorifelatious 
of  diverse  abilities^  are  ^nejtherJbigU 

GENIUS  357 

should  expect  all  degrees  of  generality  of 
.genius.     A  .few  individuals  may  excel  highly  in  a  large 
its  _.  and  thus^apgear^to,ybe  all-around  gen- 

ius^s.  Some  will  excel  in  only  a  few,  traits?-  and  still  others 
may  haye  ,a  single  talent  which-  is,  sufficiently  pronounced 
tQjpiLt.lhera  in  the  category  of,  genius.  There  is  no  neces- 
sary relationship,  either  positive  or  negative,  between 
excellence  in  one  trait  and  in  any  other. 


Theories  on  the  nature  and  causes  of  genius  are  legion, 
The  genius  has  been  credited  with  a  wide  variety  of 
attributes,  ranging  from  divine  inspiration  and  a  super- 
human "spark"  to  imbecility  and  insanity.  Among  these 
diverse  theories  it  is  possible  to  discern  four  underlying 
viewpoints.  These  will  be  designated  the  pathological, 
psychoanalytic,  typological,  and  deviationaL1 

Pathological  tjieoriesj^ 

.degeneracy!  and  even  feeblemindedness.    This  view  was 

*SW»»3»F*<»»»^^  *"         '"          "  '  "    "  """'!.'   -"i" 

put  forth  by  many  of  the  ancient  writers,  among  whom 
may  be  mentioned  Democritus  and  Seneca.  More  re- 
cently, it  received  clear  expression  in  the  writings  of 
claimed  that  all  genius  is  a  neurosis,  and 


out,  is  equally  cha^tejj^ji^^ 

EssenHallythe  same  theory  was  later  expanded  by  Nisbet 
(24)  in  England,  who  held  genius  to  be  a  mere  sport  of 
nature  or  spontaneous  variation  which  betrays  a  lack  of 
adaptation  to  its  environment. 

The  greatest  impetus  to  the  development  of  the  patho- 
logical view*4of^f  Sinus  "was  probably  furnished  by  the 

1  These  terms  are  only  employed  as  convenient  designations  and  are  not  to  be 
regarded  as  a  sufficient  characterization  of  any  of  the  views.  The  terms  "typo- 
logical" and  "deviational,"  for  example,  should  be  accompanied  by  qualifying 
adjectives,  but  the  latter  are  omitted  for  the  sake  of  conciseness. 


Italian  anthropologist,  Lombroso  (22).  His  book  entitled 
The  'Man  of  Genius  was  translated  into  several  languages 
and  read  widely  at  the  turn  of  the  present  century.  In  it 
he  attributed  to  the  genius  a  wide  variety  of  qualifies 
which  he  regarded  as  indicative  of  species  degenerajipn. 
Among  them  are  included  short  stature,  rickets,  excessive 
pallor,  emaciation,  stammering,  left-handedness,  delayed 
development,  and  originality!  He  also  claimed  to  have 
observed  a  certain  similarity  between  the  creative  act  of 
genius  and  the  typical  epileptic  seizure. 

Havelpck  Ellis  (12)  proposed  a, two-fold,  theory  which 
Ttc^iiZQ^^c^ceptional  talents  of  genius  but ^at  the 
same  time  includes  a  certain  essential  deficiencyjin  the 
complete  picture.  This  deficiency  he  regards  as  some  in- 
nate organic  inaptitude  which  prevents  complete  adjust- 
ment to  the  activities  of  everyday  life.  He  cites  as  exam- 
ples motor  incoordination,  narrow  specialization  with 
attendant  inferiority  in  other  fields,  impracticality,  and 
similar  deficiencies  occasionally  found  in  men  of  genius.1 

The  evidence  presented  in  support  of  these  views  con- 
sists of  illustrative  examples  of  geniuses  who  manifested 
the  alleged  traits.  Such  evidence  is  open  to  all  the  criti- 
cisms of  selected  cases.  A  few  individuals  can  be  found  to 
illustrate  almost  any  theory.  Only  the  objective  observa- 
tion of  large  numbers  of  unselected  cases  can  furnish  an 
adequate  test  of  these  assumptions.  This  type  of  evidence 
will  be  presented  in  a  subsequent  section. 

Several  factors  can  be  cited  which  might  erroneously 
suggest  a  linkage  between  insanity  and  genius.  Thus 
many  geniuses  of  high  degree  may  become  maladjusted 
In  a  society  built  up  around  the  average  man  and  his 

1  This  theory  might  also  be  classified  under  the  psychoanalytic  viewpoints, 
owing  to  a  certain  resemblance  to  the  Adlerian  concept  of  organic  inferiority  in 
genius  (cf.  below). 

GENIUS  359 

needs.  This  is  particularly  noticeable  in  the  case  of  a 
very  superior  child  placed  in  a  class  of  mediocre  school- 
children. It  is  equally  true  of  superior  adults.  Geniuses, 
furthermore,  are  often  regarded  as  pathological  by  their 
fellow-men  until  the  practical  benefits  of  their  work  be- 
come tangible.  Their  achievements  are  ridiculed  until 
they  succeed.  The  familiar  example  of  Fulton  and  his 
steamboat  is  a  case  in  point.  Occasionally,  the  genius 
has  met  with  systematic  and  even  violent  opposition 
from  the  church  or  other  organized  bodies.  Life  under 
such  conditions  was  not  very  conducive  to  the  develop- 
ment of  a  stable  and  well-adjusted  personality.  Finally, 
it  may  be  added  that  even  when  the  genius  is  recognized 
and  acclaimed  as  such,  he  is  likely  to  be  surrounded  by 
such  a  blare  of  publicity  that  all  his  actions  and  idiosyn- 
cracies  become  common  knowledge.  As  a  result,  any 
behavioral  deviation  too  slight  to  attract  attention  in  a 
less  outstanding  individual  is  pounced  upon,  discussed, 
and  elaborated  until  it  may  assume  the  proportions  of  a 
neurotic  or  psychotic  symptom. 

of  .genius  emphasise  .motiva- 

tion^^ intellectual  characteristics.    The  psy- 

choanalytic concepts  o^^^sublimation  and  compensation 
are  brougj^j^j^acoount  foT^e^rcmSlaQe,  achievements 
ofgenius.  Freud  (cf.  10)  regards  the  manifestations  of 
genius  as  a  sublimation  or  substitute  outlet  for  a  thwarted 
sex  drive.  He  does,  however,  recognize  the  need  for  a 
certain  degree  of  talent  together  with  strong  motivational 
factors.  Adler  (i)  regards  the  exceptional  accomplish- 
ments of  genius  as  compensations  for  other  inferiorities, 
usually  of  an  organic  nature.  A  favorite  example  is  that 
of  great  orators,  such  as  Demosthenes,  who  developed 
their  talent  by  overcompensating  for  an  initial  habit  of 
stammering  or  some  similar  speech  defect. 


The  role  of  motivational  factors  in  the  development  of 
genius  cannot  be  ignored.  Under  certain  conditions,  a 
slight  physical  handicap  may  spur  the  individual  on  to 
greater  efforts  and  thereby  contribute  indirectly  to  his 
success.  Similarly,  thwarted  desires  or  emotional  dis- 
appointments may  cause  some  individuals  to  turn  to 
intellectual  pursuits  as  a  consolation.  Restriction  of  one's 
range  of  activities  by  physical  defects,  unpopularity,  or 
other  conditions  may  also  lead  to  more  intensive  work 
and  subsequent  success  along  a  particular  line.  These 
various  motivational  influences  are  very  simple  and  clearly 
observable  facts.  The  psychoanalytic  theories,  on  the 
other  hand,  are  burdened  with  vague  concepts,  fantastical 
allusions  to  Unconscious  forces,  and  mystifying  state- 
ments. The  simple  issues  involved  are  greatly  confused 
by  such  interpretations.  These  theories,  furthermore, 
could  only  account  for  a  relatively  small  number  of  gen- 
iuses. Nor  do  they  offer  a  satisfactory  interpretation  of 
the  essential  nature  of  genius. 

Those  views  which  we  have  classified  under  the  heading 
of  typological  r^ardwwg^ius  asja  distinct  type  differing 
cfSsX^^S^imm  thei>(restw^  Such  views 

are  to  be  distinguished  from  the  pathological  and  the 
psychoanalytic  in  that  they  regard  the  man  of  genius  as 
essentially  superior  to  the  norm.  No  inferiorities  of  any 
sort  are  implicit  in  this  concept.  The  achievements  of 
genius,  according  to  these  theories,  result  from  some 
process  or  condition  which  is  entirely  absent  in  the  ordi- 
nary man.  Such  current  expressions  as  "the  spark  of 
genius  "  reflect  the  popular  influence  of  this  point  of  view. 

The  typological  approach,  like  the  pathological,  has  a 
long  history.1  In  the  ancient  world,  genius  was  frequently 
attributed  to  divine  inspiration.  The  Greeks  spoke  of  a 

1  Cf.  17  and  29  for  references. 

GENIUS  361 

man's  "daemon"  which  was  supposed  to  possess  divine 
powers  and  to  furnish  the  inspiration  for  his  creative 
work.  Among  those  who  discussed  genius  In  these  terms 
may  be  mentioned  Plato  and  Socrates.  Christianity  ad- 
vanced the  view  that  genius  was  the  inspiration  of  a 
selected  mortal  by  God.  Or,  in  some  cases,  it  was  Satan 
who  provided  the  creative  talents,  especially  when  the 
inspiration  was  of  a  kind  contrary  to  the  dogmas  of  the 

Qualitative  distinctions  are  also  common  in  literary 
and  philosophical  writings  on  the  subject  of  genius.  Mys- 
tic insights  and  unconscious  intuitions  have  been  at- 
tributed to  the  man  of  genius.  In  this  connection  may 
be  mentioned  the  views  of  Schopenhauer,  Carlisle,  and 

In  psychological  discussions  of  genius,  this  point  of 
view  is  much  less  common.  rG.  Stanley  Hall's  proposal 
(16,  p.  320)  ^^.th^sh^^^^^^'^'S^^  are.  similar 
to  those  of ,  tjae,  jidqiescent  may  be  regarded  as  a  qualita- 
tive:' distinction.  Recently  Hirsch  (17)  has  put  forth  a 
distinctly  qualitative  view.  He  attempts  to  differentiate 
three  "dimensions"  of  intelligence.  The  first  dimension 
is  perceptual  and  cognitive  and  is  shared  by  man  and  the 
lower  animals;  the  second  is  conceptual  and  is  common 
to  all  of  mankind;  the  third  he  designates  "creative 
intelligence"  and  attributes  only  to  genius. 

Qualitative  distinctions  appeal  to  the  imagination  of 
the  public.  The  genius  whom  the  layman  acclaims  differs 
so  greatly  from  the  rest  of  mankind  in  his  achievements 
that  he  seems  to  belong  to  another  species.  A  careful 
analysis  of  the  individual's  abilities,  however,  will  reveal 
no  essentially  new  process.  And  only  a  brief  unbiased 
search  discloses  the  presence  of  intermediate  degrees  of 
capacity  in  all  lines. 


The  deviational  view  regards  genius  only  as  the  upper 
extreme  of  the  normal  distribution  curve.  The  special 
"gifts"  and  " creative  powers"  of  genius  are  possessed, 
to  a  much  lesser  degree,,  by  all  individuals.  Distinctions 
are  quantitative  and  not  qualitative.  Genius  is  defined 
in  terms  of  concrete,  measurable  behavior  rather  than  in 
terms  of  mystical  entities.  To  be  sure,  the  accomplish- 
ments of  genius  are  not  attributed  to  any  single  talent, 
but  to  a  favorable  combination  of  several  intellectual, 
motivational,  and  emotional  factors. 

It  follows  from  the  deviational  view  that  the  origin  of 
genius  is  to  be  understood  in  the  same  terms  as  that  of  all 
individual  differences.  Many  writers  on  the  subject  of 
genius,  such  as  Galton  (13,  14),  Terman  (31,  32),  Cox  (8), 
L.  S.  Hollingworth  (18),  and  Pintner  (26,  Ch.  15),  lay 
the  major  emphasis  upon  hereditary  factors.  The  ob- 
servation that  genius  runs  in  families  has  probably  given 
the  greatest  impetus  to  this  view.  In  such  cases  the  po- 
tent environmental  influence  exercised  by  family  tradi- 
tions and  family  contacts  is  overlooked.  For  an  analysis 
of  the  operation  of  hereditary  and  environmental  factors 
in  the  development  of  individual  variations  in  general, 
the  reader  is  referred  to  Chapters  III,  IV,  and  V. 


Psychological  investigations  on  the  nature  and  develop- 
ment of  genius  have  followed  a  variety  of  procedures. 
These  may  be  classified  into:  (i)  biographical  analysis, 
(2)  case  study,  (3)  statistical  survey,  (4)  historiometry, 
and  (5)  mental  measurement.  In  biographical  studies, 
all  available  printed  material  on  a  given  individual  is 
examined  in  the  effort  to  arrive  at  an  understanding  of 
the  nature  and  origin  of  his  genius.  The  investigation  is 
limited  to  a  single  individual,  who  is  usually  chosen  from 

GENIUS  363 

the  great  men  of  the  past.  This  method  has  been  em- 
ployed chiefly  by  psychoanalysts,  but  it  is  also  occasion- 
ally used  by  psychologists. 

The  case  study  method  consists  of  direct  and  controlled 
observations  of  a  single  living  individual.  Because  of  the 
difficulty  of  subjecting  adult  geniuses  to  such  an  investiga- 
tion, this  method  has  been  applied  almost  exclusively  to 
gifted  children.  Several  such  studies  on  contemporary 
child  prodigies,  and  especially  juvenile  authors,  have  been 
initiated  by  psychologists. 

The  statistical  survey  method,  like  the  biographical,  is 
based  upon  an  analysis  of  printed  records.  It  differs 
from  the  latter  method,  however,  in  several  essential 
respects.  The  purpose  of  statistical  surveys  of  genius  is 
to  discover  general  trends  in  a  large  group,  rather  than  to 
make  an  exhaustive  analysis  of  a  single  case.  All  avail- 
able information  on  a  large  number  of  men  is  obtained 
from  biographical  directories,  encyclopedias,  Who's  Who, 
and  similar  sources.  This  material  is  occasionally  sup- 
plemented from  biographies.  But  the  former  sources  are 
employed  predominantly  because  of  the  more  objective, 
reliable,  and  standardized  nature  of  their  data.  It  will 
be  noted  that  in  this  method  the  criterion  of  genius  is 
chiefly  eminence. 

The  historiometry  method  makes  use  of  all  historical 
material  on  an  individual  or  a  group  of  individuals.  The 
data  are  culled  from  a  variety  of  sources,  including  bi- 
ographies, directories,  and  original  documents  such  as 
letters  and  diaries.  The  attempt  is  made  to  obtain  as 
complete  information  as  possible,  especially  on  the  child- 
hood accomplishments  of  the  great  man.  This  material 
is  then  evaluated  in  terms  of  a  more  or  less  constant 
standard  in  order  to  arrive  at  an  estimate  of  the  indi- 
vidual's traits.  This  method  was  employed  by  Woods  (34) 


in  his  study  of  mental  and  moral  heredity  in  royalty. 
Terman  (28)  subsequently  suggested  an  adaptation  of 
historiometry  whereby  the  recorded  achievements  are 
evaluated  in  terms  of  mental  test  standards  and  norms 
of  performance  for  each  age.  From  this,  an  estimate  is 
obtained  of  the  individual's  I.Q.  in  childhood.  Terman 
estimated  by  this  method,  for  example,  that  the  I.Q. 
of  Francis  Galton  was  approximately  200. 

The  mental  measurement  method  consists  of  the  direct 
study  of  large  groups  of  intellectually  superior  children 
by  means  of  mental  tests.  Extensive  use  is  now  being 
made  of  this  method.  The  subjects  are  originally  selected 
on  the  basis  of  intelligence  test  performance,  and  subse- 
quent analyses  are  made  with  the  aid  of  standardized 
intellectual,  scholastic,  and  personality  tests. 

Each  of  these  procedures  contributes  something  which 
is  lacking  in  the  others.  Similarly,  each  by  itself  suffers 
from  special  weaknesses.  The  statistical  survey,  histori- 
ometry, and  mental  measurement  methods  can  be  applied 
to  large  groups,  and  hence  disclose  general  trends.  They 
are  also  relatively  free  from  selectional  bias.  The  bio- 
graphical and  case  study  methods,  on  the  other  hand, 
give  a  more  complete  picture  of  the  individual  and  enable 
one  to  note  the  specific  action  of  various  conditions  upon 
the  subject's  development.  The  study  of  contemporary 
living  geniuses  makes  direct  observation  possible  and 
avoids  the  judgment  errors  and  other  inaccuracies  which 
are  inevitably  present  in  historical  material.  On  the 
other  hand,  carefully  controlled  observation  on  living 
geniuses  offers  many  practical  difficulties.  A  further  dis- 
advantage in  the  study  of  contemporaries  is  the  possibil- 
ity that  the  eminence  of  some  may  be  short-lived  and 
spurious  and  that  others  who  are  laboring  in  obscurity 
may  be  recognized  as  geniuses  by  posterity. 

GENIUS  365 

Finally,  the  relative  advantages  of  studying  adult  gen- 
iuses and  gifted  children  may  be  considered.  To  investi- 
gate intellectually  superior  children  in  the  effort  to  dis- 
cover the  characteristics  of  adult  geniuses  seems  somewhat 
indirect.  Only  a  small  number  of  such  children,  further- 
more,, will  develop  into  adults  who  can  be  classified  as 
geniuses.  Children,  however,  are  available  for  prolonged 
and  controlled  observation  and  testing  which  would  be 
practically  impossible  with  adults.  A  further  advantage 
of  the  study  of  gifted  children  is  that  it  offers  a  genetic 
approach  to  the  problem.  Such  an  analysis  may  go  far 
towards  clarifying  the  origin  and  nature  of  genius. 


Investigations  on  genius  through  statistical  surveys  of 
printed  records  have  been  conducted  in  England  by  Gal- 
ton  1  (13,  14)  and  Ellis  (12),  in  France  by  de  Candolle  (9), 
Jacoby  (21),  and  Odin  (25),  and  in  America  by  Cattell 
(4,  5,  6),  Brimhall  (2),  and  Clarke  (7).  Castle  (3)  con- 
ducted a  similar  survey  on  eminent  women  of  all  countries, 
but  the  data  of  this  study  are  extremely  tentative  and 
quite  difficult  to  interpret.  The  major  findings  of  these 
studies  will  be  summarized  briefly. 

Zte^£0g/^  has 

been  generaUyJjjj^^  the 

entire"  population.  Thus  in  Cattell's  analysis  of  the  oc- 
cupations of  the  fathers  of  American  men  of  science  (5), 
the  percentage  engaged  in  the  professions  was  far  in 
excess  of  that  in  the  general  population.  The  relevant 
data  are  reproduced  in  the  following  table.  C^S^^^mr 

-  "  Ws  B3I  co^clwc 
ofjoiuLon^qf .science  ppme  from 

1  Cf.  Ch.  IV  for  a  discussion  of  specific  procedure  and  results. 

366                DIFFERENTIAL  PSYCHOLOGY 

Occupational  Group 

Percentage  in  Each  Occupational  Group 

Fathers  of  American       General  Population 
Men  of  Science                     in  1850  l 

Manufacture  and  trade 





population  most  favorably  situated  to  produce  them%  The 
son  of  a  successful  professional  man  is  fifty  limes^asjikely 
to  become  a  leading  scientific  man  as  a  boy  .taken  at 
random  from  the  community"  (5,  p.  511). 

A  similar  distribution  of  occupational  level  is  to  be 
found  among  the  eminent  men  and  women  surveyed  by 
Ellis  (12).  In  Castle's  study  (3)  of  eminent  women  of 
all  times  and  nationalities,  it  was  reported  that  33.1% 
had  fathers  in  the  "learned  professions."  ^Cox  (8,  p.  37) 
obtained  the  following  distribution  of  paternal pccupation 
in  a  group  of  282  eminent  men  and  women  Qf.a^^ountries. 

Occupational  Level  Percentage 

1.  Professional  and  nobilit7  S2-S 

2.  Semi-professional,  higher  business,  and  gentry  28.7 

3.  Skilled  workmen  and  lower  business  13 -1 

4.  Semi-skilled  3*9 

5.  Unskilled  i-i 
No  record  0.7 

The  number  of  eminent  relatives  is  also  of  interest  in 
this   connection.      It  will  be  recalled  that  in  ^alton's 
studjr  (13)^  the  ,97^eoune^ 
of  73 9  known  rej^^ 

It_mll  alsp  , be  recalled,  thaty  in.  general,,, the, .ctasfiT  the 
degree  of  relationship,  ^the  more  numerous,  ."weie^lhc^nii- 
ne$t  relatives^  Similar  results  were  obtained  in  Brim- 
hairs  investigation  (2)  of  family  resemblance  among 
American  men  of  science. 

1  Corresponding  roughly  to  the  time  when  the  fathers  of  the  scientific  men  had 
pursued  their  vocations. 

GENIUS  367 

Rather  suggestive  evidence  for  an  environmental  in- 
terpretation of  the  development  of  genius  is  furnished 
by  Cattell's  analysis  of  the  place  of  birth  of  American  men 
ofjaence  (cf.  4,  6).  In  his  1906  report^  Cattell  pointed 
out  that  cities  contributed  a  much  greater  proportion  of 
men  of  science  than  did  rural  sections.  Although  at  that 
time  the  urban  population  was  about  one-sixth  of  the 
rural  population,  it  produced  a  quarter  of  the  scientific 
men.  Even  more  striking  is  the  comparison  of  different 
states  which  varied  widely  in  their  educational  facilities. 
Below  are  shown  the  relative  number  of  scientists  born 
in  each  of  nine  states.  The  latter  have  been  chosen  as  the 
clearest  examples  of  a  definite  trend  which  has  been 
operative  during  the  last  three  decades.  Corresponding 
figures  are  shown  for  the  original  group  of  1000  scientists 
selected  in  the  year  1903  and  for  the  group  of  250  elected 
in  1932.  All  figures  have  been  expressed  in  terms  of 
1000  entries  to  permit  direct  comparison  (cf.  6,  p.  1265). 

Number  of  Cases 

Place  of  Birth  (in  terms  of  1000  entries) 

1903  1932 







New  York 





















These  dajtgjsuu^ 

m^SJ^everal  conclusions  which  are 

Lit  b¥-tJie,^QmDtete  results^loTSI^^S^^SLfflMEtr; 

(cf.  6).    lujAe  fii^dafis.  there  are  marked  discrepancies 

^-~-  ————•—'          "^^^^miMii^^  MAM***  . 



in  birthplace  correspond  very  closely  to  differences  in 
educational  opportunities  in  various  sections  of  the  coun- 
try. Thirdly,  as  educational  facilities  change,,  the  fre- 
quency of  scientists  shows  a  corresponding  change.  In 
recent  years,  for  example,  there  has  been  a  phenomenal 
development  of  education  in  the  mid-western  states.  The 
relative  quality  of  education  in  such  states  has  improved, 
new  universities  have  been  established,  the  number  of 
students  in  institutions  of  higher  learning  has  increased 
rapidly,1  and  a  powerful  tradition  has  been  built  up  which 
fosters  intellectual  activity.  These  factors  probably  ac- 
count for  the  definite  rise  in  the  number  of  scientists  from 
mid-western  states,  as  compared  to  the  New  England 
states  which  were  formerly  the  undisputed  center  of  the 
intellectual  life  of  the  country. 

Of  interest  in  connection  with  the  pathplQglcal.,theories 
ofjsjemus Js  the  relative  frequency  of  insanity  among  the 
relatiY,ei^Ql  eminent  .men,  43  wll,  as  .among  .the ..  siibj  ects 
themselves.  In  all  well  controlled  studies  in  which  the 
cases  were  not  selected  to  prove  a  point,  the  incidence  of 
intellectual  and  emotional  disorders  has  been  found  td*  be 
consistently  smaller  among  eminent  men  and  their  families 
thanjL^^  In  the  group  investigated 

by  Ellis  "(12,  p,  192)3  less  than  2%  were  reported  to  have 
had  either  insane  parents  or  insane  offspring.  Among  the 
eminent  individuals  themselves,  Ellis  mentions  44  cases  of 
emotional  disorder  out  of  a  total  group  of  1030.  Of  these, 
only  13  could  be  definitely  classed  as  insane  during  the  ac- 
tive period  of  their  lives;  19  were  either  insane  for  a  short 
period  or  manifested  very  mild  disorders;  and  12  developed 
senile  dementia  in  old  age  (cf.  12,  pp.  189-190). 

1  Cf.,  e.g.,  Eells'  analysis  of  the  "center  of  population"  of  higher  education 
from.  1790  to  1920,  which  showed  a  westward  movement  at  the  rate  of  60  miles 
per  decade  (n). 

GENIUS  369 

Other  facts  which  have  been  brought  to  light  by  these 
surveys  relate  to  age  of  pjzrents  at,  the.,  timt.Q/^birth  of  the 
child,  order  of  birth,  and  similar  "vital  statistics.  "  It  has 
been  proposed,  for  example,  that  older  parents  have  in- 
tellectually superior  children  (cf.  27).  From  a  somewhat 
different  angle,  Lombroso  (22)  claimed  that  geniuses  are 
the  offspring  of  aged  parents  and  offered  this  as  further 
evidence  of  the  pathological  nature  of  genius.  The  data 
on  this  question  are  difficult  to  interpret  because  of  the 
complicating  factor  of  social  level.  People  in  the  higher 
social  classes,  from  which  geniuses  are  most  frequently 
recruited,  tend  to  marry  later  and  therefore  have  children 
at  a  later  age.  For  this  reason  average  ages  are  in  them- 
selves inconclusive.  Among  American  men  of  science, 
III)  found  35  years  to  be  the  average  agp  of 
pi,  A£,  sub  j  ect'  ,&  birth.  For  English 

men  of  science,  Galton  (14)  found  the  corresponding 
figure  to  be  36  years.  Ellis  (12)  gives  37.1  years  for  his 
group  of  British  men  and  women  of  eminence.  In  all  of 
these  groups,  however,  the  range  of  parental  ages  at  birth 
is  extremely  wide.  In  the  majority  of  cases  the  parents 
were  in  the  prime  of  life  at  the  time  of  the  subject's  birth, 
contrary  to  Lombroso'  s  contention. 

Somewhat  more  conclusive  is  the  analysis  of  order  of 
birthjw^^  the  'eminent  individual 

is  most  oftenjJie  oldest^r  first-born  child  in  the  family. 
NeS^  comes  the  youngest  child, 

intermediate  children  having  tEeleast  ^aiTS'orBecGmmg 
eminent  (cf.  12,  35).  These  findings  are  in  direct  contra- 
diction to  the  proposed  theory  that  older  parents  have 
intellectually  more  gifted  offspring.  It  would  seem  that, 
within  the  same  family,  the  superior  child  is  most  likely 
to  be  born  when  the  parents  are  younger.  This  finding 
may  be  interpreted  in  cultural  terms.  The  first  born  has 


traditionally  enjoyed  privileges  in  our  society  that  his 
younger  siblings  may  not  have  had.  More  is  usually 
expected  of  the  oldest  son.  If  a  choice  must  be  made  for 
economic  reasons,  the  oldest  child  is  usually  allowed  to 
complete  his  education,  in  preference  to  the  younger 
children.  These  conditions  might  be  sufficient  to  produce 
a  slight  degree  of  relationship  between  birth  order  and 


The  childhood  of  great  men,  viewed  retrospectively, 
has  been  the  source  of  much  controversial  discussion. 
There  is  a  popular,  belief  l  that  many  geniuses  ;were  dulHn 
childEood.  Several  examples  are  cited  in  support  of  this 
contention.  Darwin  was  considered  by  his  teachers  to 
be  below  average  in  intellect.  Newton  was  at  the  bottom 
of  his  class.  Heine  was  an  academic  failure,  revolting 
against  the  traditional  formalism  of  the  schools  of  his 
time.  Pasteur,  Hume,  von  Humboldt,  and  other  equally 
famous  men  were  unsuccessful  in  their  school  work. 

An  examination  of  the  ay  arable,,  biographical  material 

in  sucE^Eai^  wasjn- 

fe"rre3  from  scholastic,  achievements  of  a  rather  narrow 

'scope.    The  intellectually  superior  child  may  be  just  as 

"maladjusted  in   school   as   the  dull  or  borderline   case. 

cMld  may  be  mjsuited 

to  tl^JiigM^^  The  monotonous 

*3rill  and  rote  memorization  which  constituted  such  a 
large  part  of  school  work  in  the  days  when  men  like  Darwin 
or  Hume  attended  school  would  prove  particularly  irk- 
some to  a  bright  child.  Darwin,  for  instance,  seems  to  have 
been  more  interested  in  his  collections  of  insects  than  in 
memorizing  Latin  declensions,  much  to  the  annoyance  of 

1  Also  proposed  by  Lombroso  (22). 

GENIUS  371 

his  teachers.  Thus  it  is  often  impossible  to  accept  the 
recorded  opinions  of  parents  or  teachers  regarding  the 
intellectual  status  of  great  men  in  childhood. 

More  accurate  information  can  be  obtained  from  fac- 
tual records  of  the  specific  behavior  of  the  individual  at 
various  ages.  An  early  attempt  to  conduct  such  an  analy- 
sis of  the  boyhood  of  great  men  was  made  by  Yoder  (35). 
Fifty  cases,  representing  a  wide  variety  of  occupations  or 
fields  of  eminence,  were  selected  from  the  great  men  of 
six  countries.  All  of  the  subjects  were  bora  in  the  eight- 
eenth or  nineteenth  centuries,  except  Newton,  Swift,  and 
Voltaire,  who  were  born  in  the  seventeenth.  In  general 
Yoder  found  that  ill  health  in  childhood  was  often  exag- 
gerated by  the  earlier  biographers  and  that  this  condition 
was  not  so  prevalent  as  is  supposed.  Feeble  or  delicate 
health  may,  however,  offer  advantages  in  some  cases  by 
stimulating  reading  and  intellectual  pursuits.  Dickens 
was  a  good  example  of  this.  In  regard  to  intellectual 
status,  Yoder  reports  that  excellent  memory  and  vivid 
imagination  were  often  exhibited  by  great  men  from 
early  childhood. 

A  very  painsta^  the 

chgd^cf^jreat  men  WM  ^Q^Ptpd,]?JnCox  (8).  JEer- 
manrs  aclajDtatipn  pj^thejh^ 

pfoyiX  "  T hrougli  the  examination  of  several  thousand 
biographical  references,  information  was  gathered  on  the 
traits  of  301  eminent  men  and  women  born  between  145° 
and  1850.  Particular  attention  was  given  to  childhood 
behavior,  such  as  age  of  learning  to  read,  letters  and 
original  compositions  which  may  have  been  preserved, 
early  interests,  etc.  Any  special  'circumstances  which 
might  have  influenced  the  subject's  development  were 
also  noted.  The  materia^^  and 



Each  investigator  estimated  the  lowest  I.Q.  compatible 
with  the  given  facts  for  every  subject,  and  the  average  of 
these  three  independent  judgments  was  taken  as  the 
final  minimum  I.Q.  estimate  for  the  given  individual. 

After  allowing  for  certain  inaccuracies  in  the  data,  Cox 
concludes  that  the  average  I.Q.  for  the  group  "is  not 
below  155  and  probably  at  least  as  high  as  165"  (8,  p.  217). 
The  estimated  minimum  I.Q.'s  ranged  approximately  from 
100  to  200.  The  same  geniuses  cited  by  Lombroso  and 
others  as  instances  of  early  mental  inferiority  were  in- 
variably found  to  give  evidence  of  high  I.Q.'s  during 
childhood.  Among  these  may  be  mentioned  Lord  Byron, 
Sir  Walter  Scott,  and  Charles  Darwin,  whose  estimated 
I.Q.'s  proved  to  be  150,  150,  and  135,  respectively.  Among 
those  receiving  I.Q.'s  above  180  were  Goethe,  John  Stuart 
Mill,  Macaulay,  Pascal,  Leibnitz,  and  Grotius. 

In  addition  to  intellectual  superiority,  Cox  emphasizes 
the^tmpoftance  of  favorable"  personality  traits  in  tjie 
development  of  genius.  Among  such  traits  she  mentions 
"persistence  of  motive  ancT  effort,  confidence  in  their 
aHHities,  and  great  strength  or  force  of  character"  (8, 
p.  '218).  Em^nt^jelatives?  superior  home  background, 
anyd_opggrtumlies,  for  education  ami  intellectual  contacts 

Thus  it  seems  that  genius  is  usually  foreshadowed  by 
the  intellectual  and  temperamental  traits  of  childhood. 
The  geniuses  of  tomorrow  are  probably  to  be  found 
among  the  gifted  children  of  today.  This,  coasisteacy  of 

SUQSQQC  iftGCQJWJis^^  id- 

eation of  the  hereditary  determination  of  genius.  It  might, 
on  Jijie^  pth^baad^  .,  -be,  regarded  simply,,  as  illusfti:aJ;TO^of 
the  major  irapQrtaixce,  of  e$rly  tnvimnjney>t*  The  nature 
and  direction  of  the  individual's  subsequent  abilities  are 
largely  determined  by  the  circumstances  of  his  early  life. 

GENIUS  373 

It  may  be  for  this  reason  that  the  achievements  of  the 
adult  are  reflected  in  the  interests  and  proclivities  of  his 


Since  geniuses  have  generally  displayed  superior  talents 
in  childhood,  a  direct  study  of  gifted  children  should 
prove  relevant.  The  traditional  or  popular  concept  of 
the  "child  prodigy"  is  that  of  a  weak,  sickly,  unsocial, 
and  narrowly  specialized  individual.  His  achievements 
are  expected  to  be  of  the  nature  of  intellectual  "stunts" 
and  to  have  little  or  no  practical  value. 

One  of  the  earliest  recorded  cases  of  such  a  child  prodigy 
is  that  of  Christian  Heinrich  Heineken,  whose  achieve- 
ments are  described  by  his  teacher  in  an  old  German  book 
published  in  1779  (cf.  26,  pp.  352-353).  At  the  age  of 
10  months  this  child  was  able  to  name  objects  in  pictures. 
Before  the  age  of  12  months  he  had  memorized  many 
stories  in  the  book  of  Moses.  At  14  months  he  knew  the 
stories  of  the  Old  and  New  Testaments.  At  4  years  of 
age  he  could  read  in  his  native  language,  had  memorized 
1500  sayings  in  Latin,  and  also  knew  French.  At  this 
time  he  was  able  to  perform  the  four  fundamental  arith- 
metic operations,  and  he  knew  the  most  important  facts 
of  geography.  His  fame  spread  throughout  Europe  and 
he  was  summoned  to  appear  before  the  King  of  Denmark. 
True  to  the  traditional  picture,  however,  Christian  Hein- 
rich was  a  sickly  child,  and  at  the  age  of  4  years-4  months 
he  died. 

Contrary  to  popular  belief,  the  case  of  Christian  Hein- 
rich is  not  typical.  As  an  example  of  a  highly  gifted  child 

1 A  discussion  of  teaching  methods  suitable  for  gifted  children  is  beyond  the 
scope  of  this  volume.  For  a  number  of  studies  on  this  problem  the  reader  is 
referred  to  the  Twenty-third  Yearbook  of  the  National  Society  for  the  Study  of 
Education  (36). 


who  developed  into  a  healthy  and  successful  adult  may 
be  mentioned  Karl  Witte  (cf.  33).  Born  in  Lochau, 
Prussia,  in  1800,  this  " child  prodigy"  lived  until  he  was 
83,  having  retained  his  excellent  intellectual  powers  to 
the  end.  Karl  was  literally  educated  from  the  cradle. 
His  father  was  convinced  of  the  efficacy  of  early  training 
and  undertook  to  prove  this  with  his  son.  The  child  was 
never  taught  "baby  talk."  All  the  games  he  played  were 
games  of  knowledge.  When  only  8  years  old,  he  read  with 
apparent  pleasure  the  original  texts  of  Homer,  Plutarch, 
Virgil,  Cicero,  Fenelon,  Florian,  Metastasio,  and  Schiller. 
He  matriculated  as  a  regular  student  at  Leipzig  at  the 
age  of  9.  Before  his  fourteenth  birthday  he  was  granted 
a  Ph.D.  Two  years  later  he  was  made  a  Doctor  of  Laws, 
being  at  the  same  time  appointed  to  the  teaching  staff 
of  the  University  of  Berlin. 

Karl  Witte's  father,  in  discussing  the  boy's  education, 
wrote : 

...  he  was  first  of  all  to  be  a  strong,  active,  and  happy 
young  man,  and  in  this,  as  everybody  knows,  I  have  suc- 
ceeded. ...  It  would  have  been  in  the  highest  degree  un- 
pleasant for  me  to  have  made  of  him  preeminently  a  Latin 
or  a  Greek  scholar  or  a  mathematician.  For  this  reason,  I 
immediately  interfered  whenever  I  thought  that  this  or  that 
language  or  science  attracted  his  attention  at  too  early  a 
time  (33,  pp.  63-64). 

Karl  seems  not  to  have  been  in  the  least  vain  or  spoiled. 
He  never  paraded  his  knowledge,  was  modest  and  un- 
pretentious, and  not  infrequently  tried  to  learn  from  his 
companions  what  they  knew  better  than  he.  He  had 
many  playmates  of  his  own  age  and  we  are  told  that: 
"He  got  along  so  well  with  them  that  they  invariably 
became  very  fond  of  him  and  nearly  always  parted  from 
him  with  tears  in  their  eyes"  (33,  p.  187). 

GENIUS  375 

Recent  case  studies  of  gifted  children  by  psychologists 
likewise  lend  no  support  to  the  view  that  such  children 
are  inferior  in  other  respects.  The  gifted  "juvenile 
author/'  Betty  Ford,  obtained  a  Stanford-Binet  LQ.  of 
1 88  when  tested  at  the  age  of  7  years-io  months  (cf.  30). 
She  ranked  high  in  all  other  intellectual  and  educational 
tests,  but  showed  a  special  interest  and  talent  for  the 
composition  of  prose  and  poetry.  This  child  was  reported 
to  be  in  excellent  health  and  free  from  physical  defects. 
She  was  found  to  be  a  year  or  so  accelerated  in  physical 

Betty  Ford's  superior  linguistic  abilities  were  apparent 
from  an  early  age.  At  19  months  she  could  express  her- 
self clearly  and  also  knew  the  alphabet.  By  her  eighth 
birthday  she  had  read  approximately  700  books,  includ- 
ing such  authors  as  Burns,  Shakespeare,  Longfellow, 
Wordsworth,  Scott,  and  Poe.  By  this  age  she  had  also 
written  over  100  poems  and  75  stories.  The  following  is  a 
specimen  of  her  literary  products,  written  at  the  age  of 
7  years-n  months  and  entitled  "Fairy  Definition": 

Fairies  are  the  fancies  of  an  imaginative  brain 
Which  wearying  of  earthly  realities  aspires  to 
Create  beings  living  only  in  thought 
Endowing  the  spirits  thus  created 
With  all  genius  for  giving  Happiness. 

A  case  which  attracted  wide  attention  a  few  years  ago 
is  that  of  a  boy  known  in  the  psychological  literature  as 

E .  When  first  tested  at  the  age  of  8  years~u  months, 

E obtained  a  mental  age  of  1 5-7,  which  gave  him  an 

LQ.  of  187.  He  also  did  well  on  all  other  tests  except 
those  involving  manual  dexterity.  He  is  reported  as 
being  strong  and  healthy,  but  not  much  inclined  to  in- 
dulge in  games  and  sports.  At  the  age  of  12  he  was  ad- 


mitted  to  Columbia  College.  On  the  Thorndike  Intelli- 
gence Examination  for  High  School  Graduates,  he  ranked 
second  among  483  competitors.  During  his  freshman  year 
at  college  all  his  academic  grades  were  B  or  better,  with 
the  exception  of  physical  education,  in  which  his  grade 
was  C.  He  is  described  as  being  a  "good  sport"  and 
getting  along  well  with  the  other  students.  He  received 
his  A.B.  degree  at  15,  being  also  admitted  to  Phi  Beta 
Kappa.  At  16  he  obtained  his  M.A.  degree,  and  at  18, 
when  the  final  report  appeared,  he  was  well  advanced 
towards  the  Ph.D.  degree.  On  the  CAVD  Intelligence 
Examination,  his  score  was  441,  which  falls  approximately 
in  the  upper  }  of  i%  of  college  graduates.  Thus  during 

the  ten-year  period  over  which  he  was  investigated,  E 

showed  no  tendency  to  drop  below  the  high  intellectual 
level  indicated  by  his  initial  LQ. 

These  cases  are  typical  examples  of  intellectually  superior 
children.  Exceptional  talents  in  childhood  4,re  not  incom- 
patible with  good  health,  physical  vigor,  longevity,  or  a  well- 
rounded  personality.  To  be  sure,  puny,  timid,  and  sickly 
"children  can  be  found  among  the  gifted,  as  among  the  intel- 
lectually normal  or  dull.  But  such  cases  are  very  few  and 
cannot  be  regarded  as  representative  of  the  group  as  a  whole. 


Mental  test  surveys  of  large  groups  of  intellectually 
superior  children  have  revealed  the  continuity  which 
exists  between  the  average  child  and  the  highly  gifted 
"prodigy. "  In  order  to  include  a  sufficiently  large  number 
of  cases  in  such  studies,  the  standard  of  selection  must  be 
lowered.  But  by  surveying  a  wider  range  of  superior 
intellect,  a  more  complete  picture  will  be  obtained.  Since 
the  rise  of  the  mental  testing  movement,  a  number  of 
studies  on  moderately  large  groups  of  superior  children 

GENIUS  377 

have  appeared  (cf.  29  for  references).  The  most  extensive 
investigations  so  far  conducted  by  the  mental  test 
method  are  those  of  Terman  and  his  associates,  reported 
in  Volume  I  of  the  Genetic  Studies  oj  Genius  (31).  Because 
of  the  more  comprehensive  nature  of  this  study  and  its 
essential  agreement  with  the  findings  of  other  investiga- 
tions, it  will  be  described  in  greater  detail. 

The  major  group  employed  in  Terman's  study  consisted 
of  643  California  schoolchildren  between  the  ages  of 
2-tan3Ti4,  whos^Stanford-Binet  LQ.'s  were  140  or  above. 
3Tlpecial  study  was  also  conducted  on  378  high  school 
students.  In  the  main  study,  the  gifted  children  were 
compared  in  each  of  a  series  of  tests  or  measures  with 
equated  control  groups  of  intellectually  normal  children. 
For  reasons  of  expediency,  different  control  groups  were 
employed  for  various  comparisons,  the  number  of  cases 
in  such  groups  ranging  from  about  600  to  800. 

The  social  and  cultural  level  of  the  gifted  group  was 
In  regard  to  paternal  occupation,  31.4% 

belonged  to  the  professional  class,  50%  to  the  semi- 
professional  or  higher  business  class,  n.8%  to  the  skilled 
labor  class,  and  less  than  7%  to  the  semi-skilled  or  un- 
skilled labor  class.  The  average  cultural  rating  of  the 
home  l  was  22.94  as  compared  to  20.78  for  unselected 
homes.  The,  ayer3g<^  school  .grade  reached  by  the  parents 
of  the  gifted  group  was  11.8  and  by  the  grandparents 
foro.  *  These  figures  may  be  compared  with  the  average 
ainount  of  schooling  of  the  native-born  White  draft  of 
the  United  States  army  in  the  World  War,  which  did  not 
quite  reach  the  seventh  grade.  It  should  be  noted  that 
the  average  age  of  the  army  draft  was  about  15  years 
lower  than  that  of  the  parents  of  the  gifted  children  and 

1  Obtained  by  the  Whittier  Scale,  which  takes  into  account  necessities,  neat- 
ness, size,  parental  condition,  and  parental  supervision. 


much  lower  than  that  of  the  grandparents.  Since  the 
amount  of  schooling  has  risen  markedly  in  recent  years, 
the  showing  made  by  the  parents  and  grandparents  of 
the  gifted  group  is  even  better  than  it  appears.  In  addi- 
tion, over  one-fourth  of  the  children  had  at  least  one 
parent  who  was  a  college  graduate.  The  number  of  emi- 
nent relatives  and  ancestors  was  also  far  in  excess  of  that 
which  would  be  expected  through  chance,  and  many  of  the 
families  had  highly  distinguished  geneplogies. 

We  may  next  consider  certain  vital  statistics  as  well  as 
medical  and  anthropometric  data  obtained  on  the  gifted 
arid"  control  groups.  As  in  the  studies  on  adult  genius,  a 
fcpEHponde»nee  -of  -first-born:  -chiid-rea..  -was  found  in  the 
gifted  group.  The  £reque$cyx>f  insanity  in  the  family  was 
'mucTi  lower  than  average.  Only  04%  of  the  parents  and 
0.3%  of  the  grandparents  and  great-grandparents  were 
reported  to  have  had  a  record  of  insanity.  The  gifted 
children  developed  at  a  rapid  rate  from  early  infancy. 
They  ^alked.oix.lhe.^v^X^ge  ^qne^mpnth  earlier  and  talked 
2|  months  earlier  than  -the,  control  groups.  The  onset  of 
puberty  was  also  somewhat  earlier  than  normal.  Physicians' 
examinations  showed  superior  health  and  relative  freedom 
from  defects  in  the  group  as  a  whole.  Similarly,  such  con- 
ditions as  "nervousness,"  stuttering,  headaches,  general 
weakness,  and  poor  nutrition  were  less  common  in  the  gifted 
than  in  the  control  groups.  In  ^ej^ht*and  wdg^physical 
and  muscular  development,  and,  strength,  the  overlapping 
of  "gifted  "and  contror^ro  ups,  ,wasu  Almost  ,  complete.  Such 
differences  as  did  occur,  however,  favored  the  gifted  group. 


were,  of  course,  Jan  ,i&  .  adyaace-of-4he-  -normal.  1    About 

1  This  was  partly  the  result  of  the  method  of  selection.  Teachers  were  asked 
to  name  the  brightest  children  in  each  class,  and  from  among  these  the  gifted 
subjects  were  chosen  by  mental  tests. 

GENIUS  379 

°f  ^e  gifted  children  were  accelerated  and  none  were 
The  administration  of  standardized  achieve- 
ment tests  in  school  subjects  revealed,  however,  that  the 
majority  of  these  ^children  had  already  mastered  the 
sliy££LJSS£i§r  fronToae  to  three  grades  above  that  in 
which  they  were  located.  Thus  in  respect  to  actual 
abilities,  tEe  gifted  child  is  often  retarded  and  not  acceler- 
ated in  school  status.  The  performance  of  the  gifted 
children  as  a  group  was  fairly  uniform  in  different  school 
subjects.  One^idedness  was  not  characteristic  of  the 
group-  The  superiority  of  the  gifted  children  was,  how- 
ever, greater  in  such  subjects  as  language  usage,  reading, 

^BdJl3^H5S«.!J.??!9^  anc*  ^east  'm  manual  training,  paint- 
ing, and  similar  abilities. 

Thejgifted  8rouP  displayed  a  wide  range  of  interests 
9S^^9l^?l^§^P9^Pr^  as  well  as  an  active  play  life. 
A  two-months  reading  record  kept  by  the  children  showed 
that  the  gifted  subjects  read  more  than  the  control  at 
all  ages.  At  9,  the  number  of  books  read  by  the  gifted 
group  was  three  times  that  of  the  control.  The  range  of 
topics  covered  was  also  wider  and  the  quality  of  the 
books  superior  in  the  gifted  group.  Similarly,  the  gifted 
children  were  more  enthusiastic,  had  more  intense  inter- 
ests in  general,  and  reported  more  hobbies  than  the  con- 
trol group.  XlQ]lectiQns  were  also  more  common  among 
the^gifted.  A  quesaoaaaije,  on  play  information  showed 

child  of  io  knows  "  more"'-about 

plays  and  games  than  the  average  chnd'of  13.  "Apart 
from  the  fact  that  the  play  interests  of  the  gifted  children 
were  more  mature  than  those  of  the  control  children  of 
their  own  age,  no  conspicuous  differences  were  found  in 
their  play  activities. 
In  character^^.d'^sonality  development,  the  rifted 

"^'"Wi^wiwwm**19*1**^  nt  Wk«f*,B.«i,»a,j^ifaiA«£_*j«i».^'i<iriM!  —  MB»*W.>M*«»»»«*'»'  »,«, 

children  were  also  found  to  be  in  advance  of  the  normal. 


istered  to  150  subjects,  the  average  score  was  practically 
identical  with  that  obtained  at  the  first  testing.  On  tests 
of  fairmindedness,  social  intelligence,  and  other  similar 
traits,  the  group  proved  to  be  markedly  superior  to  the 
norms.  Leadership  and  ability  to  cooperate  in  social 
enterprises  were  manifested  more  frequently  in  the  gifted 
group  than  among  normal  subjects.  Terman  reports 
that:  "The  gifted  subjects  take  part  in  a  wide  variety 
of  extra-curricular  activities  and  are  as  likely  to  gain 
recognition  in  any  one  of  several  kinds  of  non-academic 
activity  as  they  are  in  scholarship"  (32,  p.  132).  Simi- 
larly 84%  of  the  boys  and  90%  of  the  girls  were  reported 
by  parents  and  teachers  as  possessing  "good  general 
health"  and  only  i%  of  each  sex  were  said  to  be  in 
"poor  health." 

From  the  evidence  examined  in,  this  .chapter,  it  is  appar- 
enj  that  there  is  no  basis  in  fact  for  many  of  the  popular 
prejudices  regarding  genius.  Insanity  and  pathological 
conditions,  rather  than  being  characteristic  of  genius,  are 
somewhat  less  common  among  the  intellectually  ablest 
than  among  the  normal.  Nor  are  ill  health,  physical 
weakness,  and  narrow  specialization  common  attributes 
of  genius.  Most  men  of  genius  display  superior  talent 
from  an  early  age;  and  conversely,  gifted  children  tend 
to  develop  into  superior  adults.  The  environmental 
conditions  of  early  childhood  have  been  found  to  exert  a 
profound  influence  upon  the  development  of  genius.  Such 
influences  may  operate  either  through  direct  intellectual 
stimulation  or  through  the  establishment  of  motivational 
tendencies  and  potent  interests  which  will  spur  the  indi- 
vidual on0  to  overcome  obstacles  and  to  reach  a  difficult 

GENIUS  383 


1.  Adler,  A.    The  Neurotic  Constitution.    N.  Y.:  Moffat  Yard, 
1917.   Pp.  456. 

2.  Brimhall,  D.  R.    "Family  Resemblances  among  American 
Men  of  Science,"  Amer.  Naturalist,  1922,  56,  504-547;  1923, 
57,  74-88,  137-152,  326-344. 

3.  Castle,  C.  S.    "A  Statistical  Study  of  Eminent  Women," 
Arch.  PsychoL,  No.  27,  1913.  Pp.  90. 

4.  Cattell,  J.  McK.    A  Statistical  Study  of  American  Men  oj 
Science:  III.  Science,  1906,  N.  S.  24,  732-742. 

5.  .     "Families  of  American  Men  of  Science,"  I:  Pop. 

Sci.  Mo.,  1915,  86,  504-515;  II:  Sa.  Mo.,  1917,  4,  248-262; 
III:  ibid.,  1917,5,368-377. 

6.  Cattell,  J.  McK.,  and  Cattell,  J.,  ed.   American  Men  of  Sci- 
ence, Fifth  edition.  N.  Y.:  Science  Press,  1933.  Pp.  1278. 

7.  Clarke,  E.  L.    "American  Men  of  Letters,"  Stud,  in  Hist., 
Econ.,  and  Public  Law,  Columbia  Univ.,  Vol.  72,  No.  168, 
1916.    Pp.  169. 

8.  Cox,  C.  M.    Genetic  Studies  of  Genius:  Vol.  II,   The  Early 
Mental  Traits  of  Three  Hundred  Geniuses.    Stanford  Univ., 
Calif.:  Stanford  Univ.  Press,  1926.  Pp.  842. 

9.  de  Candolle,  A.    Histoire  des  sciences  et  des  savants  depuis 
deux  siecles.  Geneve:  Georg,  1873.  Pp.  482. 

10.  Dooley,  L.  "Psychoanalytic  Studies  of  Genius,"  Amer.  /. 
Psychol.,  1916,  27,  363-417. 

n.  Eells,  W.  C.  "The  Center  of  Population  of  Higher  Educa- 
tion," Sch.  and  Soc.,  1926,  24,  339-344. 

12.  Ellis,  H.    A  Study  of  British  Genius.    London:  Hurst  and 
Blackett,  1904.  Pp.  300. 

13.  Galton,  F.     Hereditary  Genius.     N.  Y.:  Macmillan,  1914. 
Pp.  379  (original  ed.,  London,  1869). 

14.  .     English  Men  of  Science.     N.  Y.:  Appleton,  1875. 

Pp.  206. 

15.  Garrison,  C.  G.,  Burke,  A.,  and  Hollingworth,  L.  S.    "The 
Psychology  of  a  Prodigious  Child,"  /.  AppL  Psychol.,  1917, 



16.  Hall,  G.  S.    Adolescence.    N.  Y.:  Appleton,  1905,  Vol.  I. 
Pp.  589. 

17.  Hirsch,  N.  D.  M.    Genius  and  Creative  Intelligence.    Cam- 
bridge, Mass.:  Sci-Art,  1931.  Pp.  339. 

1 8.  Hollingworth,   L.   S.      Gifted  Children:    Their  Nature  and 
Nurture.  N.  Y.:  Macmillan,  1926.  Pp.  374. 

19.  .     "Subsequent  History  of  E — ;  Ten  Years  after  the 

Initial  Report,"  /.  AppL  Psychol.,  1927,  1 1,  385-390. 

20.  Hollingworth,  L.  S.,  Garrison,  C.  G.,  and  Burke,  A.   "Sub- 
sequent History  of  E — ;  Five  Years  after  the  Initial  Re- 
port," /.  AppL  PsychoLy  1922,  6,  205-210. 

21.  Jacoby,  P.  Etudes  sur  la  selection  chez  I' homme.  Paris  rAlcan, 
1904  (original  ed.  1881).  Pp.  620. 

22.  Lombroso,  C.    The  Man  of  Genius.   N.  Y.:  Scribners,  1895. 
Pp.  370. 

23.  Moreau,  J.    La  psychologie  morbide.    Paris:  Masson,  1859. 
Pp.  576. 

24.  Nisbet,  J.  F.   The  Insanity  of  Genius.  N.  Y.:  Scribners,  1912 
(original  ed.,  London,  1891).  Pp.  341. 

25.  Odin,  A.    Genese  des  grands  homines,  gens  de  lettres  Frangais 
modernes.    Vol.  I,  Paris:  Librarie  Universitaire,  H.  Welter, 
ed.,  1895.    Pp.  640.    Vol.  II,  Lausanne:  H.  Mignot,  ed., 
1895-  PP- 378. 

26.  Pintner,   R.      Intelligence    Testing:  Methods   and  Results. 
N.Y.:  Holt,  1931.  Pp.  555- 

27.  Redfield,  C.  L.    Great  Men  and  How  They  Are  Produced. 
Chicago,  1915.  Pp.  32. 

28.  Terman,  L.  M.      "The  Intelligence  Quotient  of  Francis 
Galton  in  Childhood,"  Amer.  /.  PsychoL,  1917,  28,  209-215. 

29.  Terman,  L.  M.,  and  Burks,  B.  S.  "The  Gifted  Child.5'  Ch.  19 
in  Handbook  of  Child  Psychology.    C.  Murchison,  ed.     Wor- 
cester, Mass.:  Clark  Univ.  Press,  1933.   Pp.  956. 

30.  Terman,  L.  M.,  and  Fenton,  J.  C.    "Preliminary  Report  of 
a  Gifted  Juvenile  Author,"  /.  AppL  PsychoL,    1921,   5, 

31.  Terman,  L.  M.,  et  al    Genetic  Studies  of  Genius:  Vol.  L 

GENIUS  385 

Mental  and  Physical  Traits  of  a  Thousand  Gifted  Children. 
Stanford  Univ.,  Calif.:  Stanford  Univ.  Press,  1925.  Pp.  648. 

32.  Terman,  L.  M..,etal.  Genetic  Studies  of  Genius:  Vol.  III.  The 
Promise  of  Youth:  Follow-Up  Studies  of  a  Thousand  Gifted 
Children.  Stanford  Univ.,  Calif.:  Stanford  Univ.  Press,  1930. 
Pp.  508. 

33.  Witte,  K.    The  Education  of  Karl  Witte  (tr.  by  L.  Wiener). 
N.  Y.:  Crowell,  1914.  Pp.  312. 

34.  Woods,  F.  A.  Mental  and  Moral  Heredity  in  Royalty.  N.  Y. : 
Holt,  1906.  Pp.  312. 

35.  Yoder,  A.  H.    "The  Study  of  the  Boyhood  of  Great  Men," 
Fed.  Sem.,  1894-96,  3,  134-156. 

36.  Twenty-third  Yearbook  of  the  National  Society  for  the  Study 
of  Education,  1924. 


Specialization  of  function  between  the  sexes  has  been 
a  powerful  social  tradition  in  almost  all  cultural  groups. 
The  particular  tasks  assigned  to  each  sex  vary  from  group 
to  group  and  are  even  occasionally  reversed,  but  some 
differentiation  of  activity  is  practically  universal.  These 
distinctions  are  impressed  upon  the  individual  from  early 
childhood,  either  by  actual  overt  differences  in  training 
and  play-activities,  or  by  the  more  subtle  but  equally 
potent  inculcation  of  traditional  beliefs  and  ideals.  It  is 
apparent  that  in  most  societies  the  effectual  environments 
of  the  two  sexes  are  fundamentally  diverse  from  an  early 
age.  Under  such  conditions,  we  should  expect  pronounced 
variation  in  the  emotional  and  intellectual  development  of 
the  two  sexes.  By  a  curious  circular  argument,  however, 
these  socially  conditioned  behavioral  differences  are  attrib- 
uted to  innate  factors. 

The  belief  in  hereditary  sex  differences  in  mental  and 
emotional  traits  is  an  old  and  persistent  one.  It  is  only 
since  the  development  of  objective  and  quantitative  test- 
ing methods  that  the  notion  of  female  inferiority  has  been 
dispelled  among  scientists.  In  the  general  public,  this 
belief  still  prevails,  as  is  manifested  by  the  reluctance  to 
open  certain  educational  and  professional  opportunities 
to  women  and  by  the  frequent  discrimination  against 
individuals  on  the  basis  of  sex  alone.  The  reasoning 
underlying  such  practices  is  that  it  would  be  futile  to  at- 
tempt to  train  men  and  women  alike,  since  the  existing 
differences  in  their  behavior  are  so  clearly  apparent.  This 



view,  of  course,  fails  to  consider  the  possibility  that  the 
existing  sex  differences  may  themselves  be  the  result 
of  the  diverse  training  and  environment  of  the  two 


The  relative  intellectual  achievements  of  men  ancJ 
women  through  the  ages  have  frequently  been  cited  a? 
evidence  of  a  sex  difference  in  ability.  An  examination  of 
any  biographical  directory  or  encyclopedia  shows  a  far 
greater  number  of  men  than  women  to  have  achieved 
eminence.  And  of  the  few  women  listed  in  such  com- 
pendiums,  many  acquired  fame  through  special  circum- 
stances rather  than  through  the  possession  of  exceptional 
talent.  In  Ellis'  study  (9)  of  British  genius,  only  55 
women  were  included  in  the  total  group  of  1030  subjects* 
Nor  did  the  standard  of  eminence  seem  to  be  higher  ioi 
women  than  for  men.  On  the  contrary,  Ellis  claims  that 
many  of  the  women  in  his  group  had  become  famous 
"on  the  strength  of  achievements  which  would  not  have 
allowed  a  man  to  play  a  similarly  large  part"  (9,  p.  10). 

Cattell,  in  a  carefully  drawn  up  list  of  the  1000  most 
eminent  persons  in  the  world,  lists  only  32  women.  Of 
these,  II  were  hereditary  sovereigns  and  8  became  emi" 
nent  through  misfortune,  beauty,  or  some  other  circum- 
stance. This  leaves  an  extremely  small  number  who  may 
be  said  to  have  distinguished  themselves  through  theif 
superior  talents  (5,  p.  375). 

Similar  results  were  obtained  by  Castle  (4)  in  her  sta- 
tistical study  of  eminent  women.  A  total  of  868  names  of 
women  were  collected,  representing  42  nations  and  cover- 
ing a  wide  range  of  epochs  from  the  seventh  century  B.C. 
to  the  nineteenth  century.  The  largest  number  of  women  in 
the  group  achieved  eminence  through  literary  pursuits,  337 


or  38.8%  of  the  subjects  being  classified  in  this  field.  The 
highest  degree  of  eminence,  however,  as  indicated  by  the 
number  of  lines  allotted  to  the  individual  in  standard 
biographical  directories,  was  obtained  by  women  as 
sovereigns,  political  leaders,  mothers  of  eminent  men,  and 
mistresses.  Among  the  other  non-intellectual  factors 
through  which  women  achieved  fame  in  the  past  are 
listed  marriage,  religion,  birth,  philanthropy,  tragic  fate, 
beauty,  and  " immortalized  in  literature." 

In  interpreting  these  results,  Castle  seems  to  recognize 
certain  environmental  influences.  Thus  she  states:  "It  is 
probable  that  woman  has  had  more  opportunity  in  litera- 
ture than  in  any  other  line  of  work.  Her  actions  have  been 
restricted  in  various  degrees  at  different  times,  and  in  dif- 
ferent localities,  and,  to  a  certain  extent,  her  thought  has 
been  regulated"  (4,  p.  41),  In  spite  of  this  assertion, 
however,  Castle  suggests  on  the  following  page  that, 
"the  common  concept  of  woman  as  a  creature  of  feeling 
rather  than  a  creature  of  reason  may  not  be  without 
foundation.  If  this  conception  is  just,  our  classification 
tends  to  show  that  when  woman  has  attained  eminence,  it 
has  not  been  in  spite  of  her  femininity,  but  rather  be- 
cause of  it"  (4,  p.  42).  Since  both  the  qualifications  for 
"eminence"  and  the  attributes  of  "femininity"  are  cul- 
turally determined,  such  a  conclusion  is  rather  redun- 

In  more  recent  times,  the  discrepancy  in  number  of 
men  and  women  who  have  distinguished  themselves  in 
intellectual  pursuits  is  still  large,  although  constantly 
diminishing.  In  the  1927  edition  of  American  Men  of 
Science  (cf.  6,  p.  1264),  there  were  listed  725  women  out 
of  a  total  of  9785  entries  in  the  pure  sciences.  The  percent- 
age of  women  in  the  various  fields  ranged  from  2.1%  in 
physics  to  22%  in  psychology.  In  the  group  of  250  scien- 


is  ts  who  were  "  starred"  *  in  the  1933  edition,  only  3  women 
vere  included  (6). 

It  is  obviously  impossible  to  draw  any  conclusions  re- 
garding the  innate  abilities  of  the  two  sexes  from  such 
:omparisons.  The  recorded  differences  in  achievement 
:ould  be  fully  accounted  for  in  terms  of  the  environmental 
renditions  which  have  prevailed.  Many  types  of  occupa- 
ions  have  been  completely  closed  to  women  until  recently. 
Thus,  on  the  basis  of  their  sex  alone,  women  have  been 
effectively  barred  from  achieving  eminence  in  a  number 
>f  fields.  When  women  have  eventually  been  admitted 
officially  to  such  vocations,  prejudice  and  discrimination 
igainst  them  have  still  been  so  prevalent  that  only  a  few 
:ould  succeed.  Even  today,  competition  is  not  on  an 
iqual  basis  for  men  and  women  in  most  occupational  fields. 

Educational  opportunities 2  have  likewise  been  very 
lissimilar  for  the  two  sexes,  although  at  present  the  en- 
vironments of  the  two  sexes  are  more  nearly  equated  in 
:his  respect  than  in  any  other.  Institutions  of  higher 
earning  were  slow  to  open  their  doors  to  women.  Al- 
:hough  America  was  in  advance  of  most  other  countries 
n  the  education  of  women,  until  nearly  the  middle  of  the 
lineteenth  century  there  was  not  a  single  institution  of 
;ollegiate  rank  in  this  country  which  admitted  women. 
Professional  and  post-graduate  education  was  not  avail- 
able until  a  much  later  date.  Even  in  the  elementary  and 
secondary  schools,  the  traditional  curriculum  of  girls  was 
different  from  that  of  boys,  including  much  less  science 
and  more  literature,  art,  and  other  "genteel"  subjects. 

Nor  can  general  home  influences  be  disregarded.    Even 

1  The  starred  men  of  science  represent  approximately  the  1000  most  eminent 
scientists  in  the  country.  The  original  1000  were  selected  in  1903;  250  additions 
were  made  in  1933,  similar  additions  having  been  made  at  successive  five-year 



in  the  most  enlightened  and  progressive  homes,  differ- 
ences are  introduced  in  the  environments  of  boys  and 
girls  which  may  prove  very  important  in  determining 
subsequent  development.  In  general,  girls  are  considered 
weaker  and  more  frail  than  boys;  they  are  sheltered  more 
and  are  taught  to  be  neater  and  quieter  than  their  brothers. 
Boys  and  girls  are  given  different  toys  to  play  with  and 
different  books  to  read.  All  of  these  apparently  minor 
environmental  factors,  operating  constantly  and  from  a 
very  early  age,  may  exert  a  lasting  influence  upon  the 
development  of  the  child's  interests,  emotional  character- 
istics, and  intellectual  talents. 

Finally,  the  relatively  intangible  but  highly  effective 
factor  of  social  expectancy  should  be  mentioned.  This 
operates  to  perpetuate  all  group  differences,  once  they 
have  been  established.  What  is  expected  of  an  individual 
is  a  powerful  element  in  the  determination  of  what  he 
will  do.  When  such  expectation  has  the  force  of  social 
tradition  behind  it  and  is  corroborated  at  every  instant 
by  family  attitudes,  everyday  contacts  in  work  and  play, 
and  nearly  all  other  encounters  with  one's  fellow-beings, 
it  is  very  difficult  not  to  succumb  to  it.  As  a  result,  the 
individual  himself  usually  becomes  convinced  that  he  is 
"superior"  or  " inferior,"  or  that  he  possesses  this  or  that 
talent,  interest,  or  attitude,  according  to  the  dictates  of 
his  particular  culture. 

During  the  last  decade  of  the..&ineteentk  ceatuxv,  the 

^pjes^g^^  *    ' 

doctrine ,  of ^j^^ife  1  be- 

came  prominent.  It  was  pointed  out  that,  although  the 
average  ability  of  men  and  women  might  be  equal,  the 

1  The  possibility  of  greater  male  variability  in  physical  traits  was  originally 
alluded  to  by  Darwin,  although  he  does  not  seem  to  have  considered  the  problem 
of  great  importance  (cf,  20). 


distribution  of  ability  in  one  sex  might  cover  a  wider 
range  than  in  the  other.  Thus  it  was  suggested  that  the 
variability  of  intelligence  among  males  is  greater  than 
among  females,  there  being  more  men  than  women  at 
either  extreme  of  the  distribution.  These  hypothetical 
distributions  are  illustrated  in  Figure  46.  It  will  be  noted 



-*  —  Measure  of  Intellect  -  *- 


that,  theoretically,  the  averages  of  two  groups  can  be 
identical  while  the  ranges  differ  considerably. 

Tlxedoctrine  of  greater  male  variability  was  regarded 

•™-™*~-»™^^  ',.,   ,«(     ,„,'"•'    '!,""  .,    '"     "    t'/'t'    '""    n.WS'ii*/*.'*,^   V""'  --1 

as  a  fundamental  biological  law  and  wa^  .believed^  t£whold 

well  £&  ^ejatalft 
of  .its,,  chief  protagonists,  wrote  as  follows: 

From  an  organic  standpoint,  therefore,  women  represent 
the  more  stable  and  conservative  element  in  evolution  (8, 
p.  421)  ...  in  men,  as  in  males  generally,  there  is  an  organic 
variational  tendency  to  diverge  from  the  average,  in  women, 
as  in  females  generally,  an  organic  tendency,  notwithstand- 
ing all  their  facility  for  minor  oscillations,  to  stability  and 
conservatism,  involving  a  diminished  individualism  and  vari- 
ability (8,  p.  425). 

This  doctrine  enjoyed  a  long  popularity  and  was  ac- 
cepted by  several  psychologists  in  their  analysis  of  sex 
differences  (cf.,  e.g.,  5,  24).  The  evidence  offered  in  sup- 
port of  the  greater  intellectual  ykriability  of  tKe  , 


two-fold.  On  the  one  hand,  the  statistics  on  eminence 
were  cited  as  proof  of  the  greater  frequency  of  superior 
intellect  as  well  as  of  the  presence  of  more  extreme  posi- 
tive deviations  among  the  male  sex.  Similar  data  were 
introduced  to  establish  the  wider  range  of  male  intelli- 
gence at  the  lower  end  of  the  distribution.  Surveys  of 
institutions  for  the  feebleminded  in  several  countries  re- 
vealed a  consistent  excess  of  males  among  the  inmates. 
Thus  it  was  argued  that  there  were  more  idiots  as  well  as 
more  geniuses  among  men,  and  that  women  as  a  group 
tended  to  cluster  more  closely  around  the  average  or 
mediocre  degrees  of  ability. 

The  cultural  basis  of  sex  differences  in  the  attainment 
of  eminence  has  already  been  discussed.  No  biological 
law  need  be  invoked  to  account  for  the  greater  frequency 
of  men  in  the  biographical  directories  and  encyclopedias. 
The  greater  incidence  of  males  in  institutions  for  mental 
defectives  has :  likewise  beeiTlound  to  result 'from  cultural 
~fa50*9»  This  was  especially  ^ggjonstrated  in  a  study  by 
1*. L  S. ,  Hollingworth  (14)  on  1000  cases  referred  for  exam- 
ination to  the  Clearing  HSuse  for  Mental  Defectives  at 
the  Post  Graduate  Hospital  in  New  York  City,  as  well  as 
1142  cases  actually  committed  at  the  New  York  City 
Children's  Hospital.1  Analysis  of  intelligence  test  scores 
and  other  available  data  revealed  the  differential  opera- 
tion of  a  selective  factor  upon  the  two  sexes. 

In  the  first  place,  the  males  referred  for  examination,  as 
well  as  those  actually  committed,  were  on  the  average 
much  younger  than  the  females.  Secondly,  the  I.Q.'s  of 
the  females  presented  for  examination  were  lower  than 
those  of  the  males.  This  difference  in  LQ.  was  even  greater 
when  the  cases  actually  committed  were  compared.  A 

1  An  institution  for  mental  defectives,  then  located  at  Randall's  Island, 


survey  of  the  previous  occupations  and  general  case  his- 
tories of  the  subjects  suggested  that  the  probable  explana- 
tion ofj:hese  findings  lies  in  the  uncompetitive  nature  of 
many  occupations  open  to  women.  ^nugjg^^^titfi..dettec- 
Tion  of  feeblemindedness  as  well  as  the  necessity  of  com- 
mitment less.likely  among  women  than  among  men.  A 
woman  of  moron  level  can  survive  outside  of  an  institu- 
tion by  turning  to  housework,  prostitution,  or  marriage  as 
a  means  of  livelihood.  The  boy,  on  the  other  hand,  is 
forced  into  industrial  work  at  a  relatively  early  age  and 
will  soon  reveal  his  mental  deficiency  in  the  severe  com- 
petition which  he  encounters.  Thus,  although  there  is  an 
excess  of  males  in  institutions  for  mental  defectives,  it 
^would  seem  that  there  are  more  feebleminded  females 
outsicte  of  institutions. 

'" "ttarl  Pearson  (20)  was  among  the  first  to  challenge  the 
adequacy  !j5Ot33535&.  s$x  differences  in 
a  comparison  of  the  extremes  of  the  distribution.  He 
called  """attention  to  the  need  for  direct  measurement  of 
variability  around  the  average  in  large  groups  of  unselected 
subjects.  Pearson  himself  computed  coefficients  of  rela- 
tive variability  for  several  classes  of  data,  consisting 
chiefly  of  physical  and  anatomical  measurements  on 
adults.  He  found  no  evidence  of  greater  male  variability, 
but  rather  a  slight  tendency  to  greater  female  variability. 
Similarly ^JHbUiogworth  and  Montague  (15)  collected  a 
krge^nu^mber  of jAysica^  0^^,jqQp"mife 

andjooa, Jemale,,,ijafaat& ,  M  ^irth^^ 

possible  effects  of  differential  environment.  No  consistent 

r, ,  .-<-'^^^^^^ 

sex  dmerence  in  variability  was  found. 

^^iri^s'^^^tS^^nSw''  available  on  male  and  female 
variability  in  a  wide  variety  of  traits.1  In  such  character- 

1  For  summaries  of  a  large  portion  of  this  material,  cf.  Henmon  and  Living* 
pAone  (13)  and  Lincoln  (17). 


istics  as  height,  weight,  physiological  maturity,  dentition, 
and  anatomical  development,  the  data  are  inconsistent. 
The  relative  variability  of  the  two  sexes  differs  with  the 
specific  trait  under  consideration,  the  age  of  the  subjects, 
their  social  and  economic  level,  and  even  the  particular 
community  in  which  the  data  are  obtained.  Intelligence 
test  results  exhibit  a  similar  lack  of  consistency.  On  indi- 
vidual tests  such  as  the  Stanford-Binet,  girls  seem  some- 
what more  variable,  whereas  boys  show  wider  variability 
on  many  group  tests.  Age  is  also  a  factor  in  determining 
the  relative  variability  of  the  sexes  on  these  tests. 
The  same  is  true  of  variability  In  school  achievement. 
The  findings  differ  with  the  specific  situation,  in  one  case 
the  boys  being  more  variable,  in  another  the  girls.  In  the 
large  majority  of  cases,  furthermore,  the  differences  in 
variability  in  favor  of  either  sex  are  too  slight  to  be  of 
much  significance. 


In  all  group  comparisons,  selective  factors  may  operate 
to  vitiate  the  results.  When  a  group  is  not  a  random  or 
representative  sample  of  the  population  from  which  it  is 
drawn,  it  is  said  to  be  a  select  group.  Such  a  sampling  is 
unsuited  for  any  type  of  investigation,  since  any  results 
obtained  with  it  could  not  be  generalized  but  would  apply 
only  to  the  specific  group  employed.  An  additional  com- 
plication in  the  comparison  of  two  populations  arises 
from  the  fact  that  selection  may  have  operated  differently 
in  the  two  groups.  Thus  if  a  group  of  college  girls  were 
compared  with  trade  school  boys,  the  two  samplings  would 
be  selected  in  different  ways.  Not  only  is  neither  group 
representative  of  men  or  women  in  general,  but  the  one 
represents  the  upper  end  of  the  female  distribution  and 
the  other  a  central  or  slightly  inferior  segment  of  the  male 


distribution.  In  addition  to  being  unrepresentative,  these 
groups  are  not  comparable. 

Selective  factors  are  often  difficult  to  detect  and  usually 
difficult  to  control.  An  example  of  a  selective  factor  whose 
presence  was  repeatedly  overlooked  is  furnished  by  the 
data  on  the  relative  incidence  of  males  and  females  in 
institutions  for  the  feebleminded  (cf.  above).  A  similar 
selective  influence  has  been  demonstrated  in  high  school 
enrollment.  It  would  seem  that  groups  of  boys  and 
girls  attending  the  same  high  school  constitute  truly 
comparable  samples  for  the  study  of  sex  differences.  But 
investigations  on  elementary  and  high  school  students 
have  demonstrated  that  this  is  not  the  case. 

In  two  separate  studies,  the  Pressey  Group  Test  of 
intelligence  was  administered  to  2544  elementary  school- 
children between  the  ages  of  8  and  16  (22)  and  to  5929 
high  school  seniors  ranging  in  age  from  16  to  23  (3).  The 
percentages  of  boys  who  reached  or  exceeded  the  median 
score  of  the  girls,  as  well  as  the  number  of  cases  in  each 
group,  are  shown  below.  In  the  elementary  school  study, 
the  data  are  reported  separately  for  each  age  group  (cf  .  22, 
p.  327).  In  the  study  on  high  school  seniors,  a  single 
summary  figure  is  given  for  the  entire  group  (cf.  3,  p.  61). 

Elementary  School  Group 


of  Cases 



















1  02 


High  School  Seniors 




<„  Exceeding  Girls'  Median 




It  will  be  noted  that  in  the  elementary  school  grades  the 
girls  excel  at  all  ages,  although  the  sex  difference  is  negli- 
gible among  the  i6-year-olds.  Among  the  high  school 
seniors,  however,  this  relationship  is  reversed,  over  50% 
of  the  boys  reaching  or  exceeding  the  girls5  median  score. 
This  reversal  becomes  intelligible  if  we  examine  the 
relative  number  of  each  sex  in  the  elementary  grades  and 
in  the  senior  year  of  high  school.  Throughout  the  high 
school  period  there  is  a  much  more  rapid  elimination  of 
boys  than  girls.  Boys  whose  academic  work  is  not  satis- 
factory are  more  likely  to  leave  school  and  go  to  work, 
whereas  girls  tend  to  be  kept  in  school  longer.  Girls 
also  seem  to  adjust  better  to  the  school  curriculum  and 
school  routine  in  general.  The  less  intelligent  girls  will 
exert  more  effort  and  manage  to  pass  sufficient  subjects  to 
stay  in  school,  while  boys  in  the  same  situation  are  more 
likely  to  rebel  against  school  work.  This  explanation  was 
borne  out  by  an  examination  of  the  scholastic  history  of 
those  students  who  had  dropped  out  in  the  course  of  their 
high  school  work.  Owing  to  the  differential  action  of  this 
selective  influence  upon  the  two  sexes,  differences  between 
the  intelligence  test  scores  of  high  school  boys  and  girls  can- 
not be  attributed  to  a  true  sex  difference.  In  the  evaluation 
of  any  study  on  group  differences,  selective  factors  are  one 
of  the  most  subtle  forms  of  error  to  be  guarded  against. 


A  further  complication  in  the  analysis  of  sex  differences 
arises  from  a  difference  in  the  developmental  rate  of  boys 
and  girls.  Itjias  been  clearly  established  that  in  physical 
characteristics  girls  reach  maturity  earlier  than  boj;s. 
Moreover,  at  any  one  age  (luring  childhood,  girls  are 
usually  farther  advanced  towards  their  ultimate  adult 
status  than  boys. 


Several  investigators  have  compared  the  height  and 
weight  of  boys  and  girls  at  successive  ages.  In  order 
directly  to  compare  the  developmental  status  of  the  two 
sexes  in  these  traits,  each  age  average  can  be  expressed 
as  a  percentage  of  the  adult  norm  for  that  sex.  In 
Table  XVI  will  be  found  such  percentages  for  boys  and 
girls  between  the  ages  of  6  and  17.  The  figures  are  based 
upon  data  from  several  investigations.  It  will  be  noted 
that  at  each  age  measured,  the  girls  have  attained  a  greater 
percentage  of  their  adult  height  and  weight  than  the  boys. 
Similar  results  were  obtained  in  an  extensive  investigation 
by  Baldwin  (2),  in  which  the  same  subjects  were  measured  at 
successive  ages.  At  certain  ages  the  developmental  acceler- 
atToonoFfKie  girls  is  so  great  that  they  are  actually  taller  &nd 
heavier  than  boys,  in  absolute  measures.  In  Baldwin's  data, 
the  girls  were  found  to  be  superior  in  height  between  the 
ages  of  ii  and  13,  and  in  weight  between  9  and  16. 



(After  Lincoln,  17,  p.  20) 



































































Other  aspects  of  physical  development  show  a  similar 
acceleration  of  the  female  sex.  It  is  a  well  known  fact 
that  girls  reach  the  age  of  reproductive  maturity  earlier 
than  boys.  In  two  investigations  (7,  i)  on  about  4800  boys 
and  1241  girls,  respectively,  the  Children  were  classified 
into  pubescent  and  post-pubescent  on  the  basis  of  several 
physiological  observations.  The  average  age  at  which  the 
girls  entered  the  post-pubescent  stage  was  between  13  and 
13!;  the  boys  reached  this  stage  approximately  one  year 
later.  Except  at  age  16,  the  percentage  of  girls  in  each 
age  group  who  were  classified  as  post-pubescent  exceeded 
the  percentage  of  boys  so  classified  by  a  substantial 


Anatomical  development  has  been  measured  by  the 
relative  degree  of  ossification  of  the  bones  of  the  hand. 
In  this  also,  girls  have  been  found  to  be  in  advance  of 
boys,  the  median  anatomic  indices  of  the  girls  being  higher 
than  those  of  the  boys  at  every  age  from  6  to  1 6  (cf.  21), 
A  similar  difference  has  been  found  in  dentition.  In  gen- 
eral, girls  shed  their  deciduous  teeth  sooner  and  get  their 
permanent  teeth  at  an  earlier  age  than  boys.  In  the  case 
of  certain  teeth,  these  differences  amount  to  one  year  or 
over  (cf.  12). 

The  significance  of  sex  differences  in  the  rate  of  physical 
growth  has  been  emphasized  by  several  writers  (cf.,  e.g.,  3, 
17,  22).  It  has  been  suggested  that  girls  might  be  acceler- 
ated in  mefttal  as  well  as  physical  development.  Thus  the 
'fact  that  girls  of  elementary  school  age  excel  on  most 
intelligence  tests  has  been  attributed  to  this  factor.  If 
this  were  the  case,  equated  age  groups  of  boys  and  girls 
would  not  be  comparable.  It  would  then  be  necessary  to 
equate  the  sexes  in  regard  to  developmental  stage  or 
physical  maturity  rather  than  chronological  age.  But 
such  a  procedure  would  introduce  an  inequality  in  amount 


of  training  and  general  environmental  stimulation.  This 
problem  only  arises,  of  course,  in  the  comparison  of  chil- 
dren, and  does  not  apply  to  adults.  Children,  however, 
have  been  the  most  frequent  subjects  for  surveys  on  sex 
differences,  both  because  of  their  greater  accessibility 
in  large  numbers  and  because  they  have  been  exposed  to  a 
relatively  more  homogeneous  environment. 

It  should  be  noted  that  mental  acceleration  of  girls 
has  lioFTTeen  directly  demonstrated.  Its  possibility  has 
only  been  inferred  by  analogy  with  physical  development. 
It  is  doubtful,  however,  whether  physical  maturity  can 
have  much  influence  upon  intellectual  development.  The 
data  on  the  relationship  of  mental  and  physical  traits  are 
too  consistently  negative  for  such  an  assumption  (cf. 
Ch.  VIII).  In  emotional  and  other  personality  traits  it  is 
probable  that  the  onset  of  puberty  and  the  relative  physio- 
logical maturity  of  the  individual  introduce  a  disturbing 
factor  in  sex  comparisons  at  certain  ages.  But  in  regard  to 
the  child's  intellectual  status,  the  environmental  stimula- 
tion to  which  he  has  been  exposed  is  far  more  significant 
than  slight  differences  in  physical  condition. 


In  the  evaluation  of  any  obtained  difference  between 
two  groups,  it  is  necessary  to  determine  the  statistical 
reliability  of  such  a  difference.  Reliability  means,  in  this 
case,  the  degree  of  consistency  among  the  results  obtained 
on  different  samplings  of  the  same  population.  The  prob- 
lem of  reliability  arises  from  the  fact  that  in  any  investi- 
gation only  a  sample  of  the  entire  population  is  employed. 
For  example,  if  the  population  under  investigation  is 
defined  as  public  school  children  in  American  cities,  data 
may  be  gathered  on  some  5000  or  6000  children  in  a  dozen 
schools.  From  these  results,  the  investigator  generalizes 


to  the  entire  population.  If  the  sampling  was  carefully 
chosen  to  be  representative  of  the  given  population,  such 
conclusions  will  not  be  far  in  error.  The  figures  thus 
obtained,  however,  will  not  be  identical  to  those  which 
would  have  been  secured  by  testing  the  entire  population 
of  American  city  public  school  children.  Nor  will  the 
results  from  successive  samplings  of  the  population  coin- 
cide perfectly.  Had  a  different  sampling  of  5000  city 
public  school  children  been  employed,  slightly  different 
results  would  have  been  obtained. 

This  variation  in  results  from  sampling  to  sampling 
within  the  same  population  is  known  as  a  sampling  error. 
Statistical  measures  of  reliability  furnish  a  theoretical 
estimate  of  the  probable  limits  within  which  such  errors 
will  fall.  Formulae  are  available  for  the  computation  of 
the  sampling  error  of  all  statistical  measures,  such  as 
averages,  differences  between  averages,  measures  of  varia- 
bility, and  correlation  coefficients.  It  is  thus  possible  to 
estimate  the  maximum  amount  of  variation  to  be  expected 
in  any  statistical  measure  if  the  experiment  were  repeated 
on  another  sampling  of  the  same  population. 

The  most  common  measures  of  reliability  are  the  "prob- 
able error"  (P.E.)  and  the  "standard  error"  (S.E.  or  <r). 
The  total  estimated  range  within  which  a  given  measure 
will  fall  in  successive  samples  is  covered  by  =b  4  X  P.E, 
and  =h  3  X  S.E.  An  hypothetical  example  will  serve  to 
illustrate  the  application  of  such  measures.  If  the  aver- 
age score  obtained  by  a  group  is  40  points  and  the  P.E. 
of  this  average  is  found  to  be  ^  points,  then 

Therefore,  in  successive  samples  of  the  population  from 
which  this  group  was  drawn,  the  average  might  fall  be- 
tween 32  and  48.  These  extreme  values  will,  however, 


be  rare,  and  in  one-half  of  the  samples  the  average  will 
probably  fall  between  38  and  42,  or  =t  I  P.E.  from  the 
obtained  value. 

Similarly,  let  us  suppose  that  a  group  of  sixth  grade 
public  school  boys  and  girls  obtained  the  following  aver- 
age scores  on  an  intelligence  test: 

Average  Score 

Girls  90 

Boys  80 

Difference  10 

Let  us  assume  further  that  the  P.E.  of  this  difference 
is  5.  In  different  samples  of  sixth  grade  public  school  chil- 
dren, we  should  therefore  expect  the  sex  difference  to  vary 
from  30  points  (10  +  [4  X  5]  =  30)  to  — 10  points 
(10  —[4X5]  =  — 10).  In  other  words,  the  relative  standing 
of  the  male  and  female  groups  would  be  reversed  in  certain 
samplings,  and  the  boys'  average  might  be  as  much  as 
10  points  higher  than  that  of  the  girls.  This  is  what  is 
meant  by  an  unreliable  difference.  The  same  conclusion  can 
be  arrived  at  more  directly  by  dividing  the  obtained  dif- 
ference by  its  P.E.  If  the  difference  is  4  or  more  times 
as  large  as  its  P.E.,  there  will  be  no  reversal  of  direction 
and  the  difference  is  said  to  be  perfectly  reliable.  In  the 
present  example,  it  will  be  noted  that  the  difference  is 
only  twice  as  large  as  its  P.E.  (10  -5-  5  =»  2).  This  value, 
called  the  critical  ratio,  is  an  index  of  the  degree  to  which 
the  obtained  figures  represent  a  reliable  or  consistent  trend. 
TJb&.<p»bA^  errpr)jof ^n  ob- 

tained , ,  difference,  4ep  wU ,  upqn^he  ^  s ize  jpT  tKeTam^lings 
employed  as  well  as  tjifi  amount  of  var^^ft^m^m^ihe 
samplings.  It  is  apparent  that  the  larger  the  sampling,  the 
more  reliably  will  the  results  be  established.  If  the  sam- 
pling were  infinitely  large,  the  probable  error  would  be 
zero,  since  the  entire  theoretical  population  would  then 


have  been  included.  In  most  of  the  earlier  investigations 
on  sex  differences,  the  samples  employed  were  so  small 
as  to  yield  extremely  large  probable  errors,  had  the  latter 
been  computed.  The  sex  differences  reported  in  such  stud- 
ies may  have  been  due  entirely  to  chance  errors  of  sam- 

Similarly,  the  wide  variability  existing  within  each  sex 
in  regard  to  any  trait  renders  the  differences  between 
averages  less  reliable.  If  all  women  were  of  identical 
height,  for  example,  and  all  men  were  likewise  equal  in 
height,  then  sex  differences  in  height  could  be  reliably 
established  by  comparing  only  one  representative  of  each 
sex.  All  other  samplings  would  yield  the  same  difference, 
since  variation  within  each  sex  would  be  zero.  The  greater 
the  variability  within  either  group,  the  larger  will  be  the 
probable  error  of  the  obtained  values.  In  the  computa- 
tion of  P.E/s  and  SJE/s,  both  the  number  of  cases  and 
the  variability  of  the  group  are  taken  into  account.1 


The  establishment  of  a  perfectly  reliable  difference 
betweOT  W6  "groups  does  i^ot  preclude  the  possibility  of  a 
large  amount  of  overlapping  between,  such  groups.  In 
Figure  47  will  be  found  the  distribution  curves  of  a  group 
of  189  boys  and  206  girls  in  the  third  and  fourth  elementary 
school  grades  on  a  test  of  arithmetic  reasoning.  The 
average  score  of  the  boys  is  40.39  and  that  of  the  girls 
35.81.  The  difference  between  the  averages  is  4.58  points 
and  the  P.E.  of  this  difference  is  only  0.58.  The  difference 
is  thus  nearly  eight  times  as  large  as  its  P.E.  and  can  be 

i  o  TT      £  S.D.  of  the  group    -n  ,-,      r 

1  S.E.  of  an  average  «  -  —  ~2  -  fT;  P.E.  of  an  average  =  .6745 


S.D.  of  the  group    «,,,..,,.  .          ,          , 

X  -  .      .  °  -  £.   For  a  further  discussion  of  the  computation  and  use  of 

these  measures,  see  Garrett  (10). 

50  r 

10-  15-  20-  25-  30-  35-  40-  45-  50-  55- 
14  19  24  29  34  39  44  49  54  59 


REASONING.   (Data  from  Schiller,  23,  p.  67.) 

regarded  as  perfectly  reliable.  An  examination  of  the 
distribution  curves,  however,  reveals  a  large  amount  of 
overlapping  between  the  two  groups.  A  very  large  per- 
centage of  boys  and  girls  fall  within  the  same  range  of 
scores.  Furthermore,  38%  of  the  girls  obtained  scores 
higher  than  the  boys'  average,  and  24%  of  the  boys  scored 
below  the  girls'  average. 

Thus,  on  account  of  overlapping,  any  relationship  estab- 
lished for  the  group  averages  will  not  necessarily  hold  for 
individual  cases.  Although  one  group  may  excel  another 
by  a  large  and  significant  amount,  individuals  can  be 
found  in  the  " inferior"  group  who  will  surpass  certain 
individuals  in  the  " superior"  group.  Owing  to  the  large 
extent  of  individual  differences  within  any  one  group  as 
contrasted  to  the  relatively  small  difference  between 
group  averages,  an  individual's  membership  in  a  given 


group  furnishes  little  or  no  information  about  his  status 
in  any  trait. 

In  the  large  majority  of  investigations  on  group  differ- 
ences, only  averages  are  considered.  For  a  complete  pip- 
ture  of  the  relative  standing  of  the  two  groups,  however, 
some  index  of  the  degree  of  overlapping  should  be  in- 
cluded. The  best  procedure  would  be  to  report  the  entire 
frequency  distributions  of  the  two  groups.  This  is  often 
impracticable,  however.  A  simpler  alternative,  in  the  case 
of  normally  distributed  samplings,  is  to  state  the  per- 
centage of  subjects  in  one  group  who  reach  or  exceed  the 
median  (or  average)  of  the  other.  Complete  overlapping 
would  then  be  indicated  if  50%  of  one  group  reached  or 
exceeded  the  median  of  the  other.1  If  more  than  50%  of 
group  A  reach  or  exceed  the  median  of  group  B,  then 
group  A  is  superior  to  group  B;  if  less  than  50%,  A  is 
inferior  to  B.  Occasionally,  some  other  value  is  substituted 
for  the  median  as  the  point  of  reference.  Thus  the  in- 
vestigator might  report  the  percentage  of  group  A  which 
reaches  or  exceeds  the  highest  score  obtained  in  group  B, 
or  the  percentage  of  group  A  which  reaches  or  exceeds 
the  upper  quarter  of  group  B. 


It  is  a  platitude  to  insist  that,  in  order  to  obtain  sig- 
nificant data  on  any  question,  an  accurate  measuring 
instrument  must  be  employed.  Yet  the  methods  of 
measurement  employed  in  the  study  of  sex  differences, 
as  well  as  in  other  group  comparisons,  have  frequently 
been  crude  and  often  wholly  unsuited  to  the  problem. 
Thus  ratings  by  associates  were  used  in  many  of  the 

1  The  curves  will  not  coincide,  of  course,  if  the  ranges  are  unequal.  In  such  a 
case,  complete  overlapping  is  obtained  only  in  the  sense  that  one  distribution  is 
contained  entirely  within  the  other. 


earlier  investigations  on  sex  differences,  and  especially 
in  those  concerned  with  personalty  traits.  Teachers' 
ratings  of  schoolchildren  were  especially  common.  It  is 
obvious  that  such  ratings  do  little  more  than  reflect  the 
systematic  bias  of  the  judges.  In  the  comparison  of  such 
groups  as  the  sexes  or  various  "races"  or  nationalities 
about  which  definite  beliefs  are  fostered  within  each  cul- 
ture, ratings  cannot  be  regarded  as  an  index  of  the  sub- 
ject's actual  standing. 

The  reliability  of  the  tests  should  also  be  considered. 
If  a  test  is  too  short  or  if  performance  on  it  is  affected  by 
too  many  irrelevant  factors,  it  will  yield  different  results 
on  repeated  administrations.  On  such  a  test,  the  scores 
of  the  same  individuals  will  vary  widely  from  time  to 
time.  These  discrepancies  in  test  scores  are  known  as 
errors  of  measurement.  Group  differences  found  with  a 
short  and  poorly  constructed  test  may  be  entirely  spurious 
and  may  be  expected  to  disappear  upon  a  reexamination 
of  the  same  subjects. 

Much  confusion  has  also  been  introduced  into  discus- 
sions of  group  differences  by  the  relatively  loose  designa- 
tions assigned  to  most  mental  tests.  If  a  test  is  labeled 
"analytic  reasoning,"  there  is  a  tendency  to  assume  that  it 
actually  measures  that  trait,  although  such  a  trait  may 
not  even  exist  as  a  unitary  function  and  may  consist  of  a 
manifold  of  independent  abilities.  Similarly,  if  two  tests 
are  given  the  same  name,  they  are  commonly  regarded  as 
measuring  the  same  function.  An  hypothetical  example 
will  show  how  this  practice  may  affect  group  comparisons. 
Let  us  suppose  that  one  investigator  has  constructed  a 
sentence  completion  test,  which  he  labels  a  measure  of 
"logical  thinking."  In  such  a  test,  as  in  most  verbal 
tests,  girls  will  probably  excel.  If  now  another  investi- 
gator also  sets  out  to  construct  a  test  of  "logical  thinking" 


and  decides  to  employ  arithmetic  problems  as  his  material, 
he  will  find  that  boys  excel  in  this  trait.  The  results  of  the 
two  studies  will  thus  seem  to  be  in  direct  contradiction, 
owing  to  the  use  of  a  common  term  to  cover  two  discrete 
types  of  mental  activity. 

Many  discrepancies  in  the  data  on  sex  differences  may 
be  attributed  simply  to  such  a  confusion  of  terminology. 
Unless  identical  tests  are  administered  in  an  identical 
manner,  we  cannot  assume  that  the  same  mental  processes 
were  measured  in  every  case.  The  use  of  a  different  time 
limit,  for  example,  might  change  a  power  test  into  a  speed 
test  and  thus  yield  entirely  different  results.  A  slight 
alteration  in  the  directions  might  make  it  more  difficult 
for  the  subjects  to  understand  what  is  required  of  them 
and  might  thereby  introduce  a  new  element  into  the  test, 
viz.,  ability  to  follow  verbal  instructions.  "Intelligence" 
scales  are  probably  the  best  example  of  the  use  of  general 
terms  in  describing  widely  diverse  tests.  Much  con- 
troversy has  been  occasioned  by  the  application  of  such 
scales.  Owing  to  the  employment  of  "intelligence"  scales 
which  sample  different  sets  of  abilities,  some  students  of 
sex  differences  conclude  that  boys  are  more  intelligent, 
others  that  girls  are  more  intelligent. 

A  closely  related  problem  pertains  to  the  use  of  "lump 
scores"  in  group  comparisons.  Group  differences  in 
specific  abilities  may  be  completely  obscured  by  the  com- 
parison of  total  or  average  scores  on  a  battery  of  tests. 
If,  for  example,  boys  excel  in  numerical  aptitude  and  girls 
in  verbal  aptitude,  and  a  scale  of  'so-called  general  in- 
telligence is  weighted  equally  with  items  from  both  fields, 
no  significant  sex  difference  in  total  score  will  be  found. 
Should  the  scale  be  overweighted  with  items  of  one  type, 
furthermore,  it  will  favor  the  group  excelling  in  that 
trait  and  indicate  an  apparent  difference  in  general  in- 


telligence.  In  recent  years,  with  the  development  of  a 
more  critical  attitude  towards  mental  testing,  there  has 
been  a  growing  tendency  to  look  for  group  differences  in 
separate  abilities  rather  than  in  " general  level  of  perform- 
ance." In  the  study  of  group  differences,  it  is  of  the  great- 
est importance  to  state  results  in  highly  specific  terms 
and  to  limit  conclusions  to  the  particular  materials, 
procedure,  and  other  conditions  of  each  investigation.1 


The  most  crucial  problem  in  any  analysis  of  group 
differences  pertains  to  their  origin.  The  theoretical  and 
practical  implications  of  behavioral  characteristics  which 
are  innately  fixed  and  of  those  which  are  socially  condi- 
tioned are  fundamentally  different.  The  question  of 
heredity  and  environment  is  always  a  difficult  one  to 
answer.  It  is  especially  elusive  in  the  case  of  human 
development  because  of  the  practical  impossibility  of 
rearing  children  under  rigidly  controlled  laboratory  condi- 
tions. And  as  was  shown  above,  we  cannot  otherwise 
assume  that  the  environments  of  the  two  sexes  have  been 
equated,  even  in  the  case  of  fraternal  twins. 

There  are,  however,  certain  indirect  sources  of  evi- 
dence which  throw  some  light  upon  the  origin  of  group 
differences.  Thus  a  biological  or  hereditary  determination 
of  sex  differences  in  ps^cfi'^  the 

existence  of  a  universal  pattern  of  male  and  Jemale  behavior. 
If,  ori  the  other  hand,  such  differences  are  environmentally . 

conditioned,  we  should  expect  the  traditional  ]behavioral 
\     ^,f8Brw^^w«^ 

characteristics  of  each  sex  to  ptfr^ JT<^^ 

D*ata   on   this    question   are    as   yet   relatively   meagre. 

Psychologists  have  generally  neglected  the  wealth  of  infor- 

1  For  a  further  discussion  of  many  of  the  difficulties  involved  in  the  study  of 
sex  differences,  see  Lehman  and  Witty  (16). 


matlon  offered  by  the  comparative  study  of  behavior 
phenomena  in  different  cultural  milieux.  Certain  facts 
collected  by  anthropologists,  however,  have  a  direct  bear- 
ing upon  this  problem. 

The  play  activities  of  boys  and  girls  have  been  a  sub- 
ject of  frequent  speculation.  Some  would  argue,  for 
instance,  that  girls  play  with  dolls  because  of  a  nascent 
" maternal  drive"  or  some  similar  innate  interest  or  emo- 
tional trait  characteristic  of  their  sex.  The  almost  com- 
plete absence  of  this  type  of  play  activity  among  boys  is 
regarded  as  an  index  of  a  fundamental  biological  diversifi- 
cation in  emotional  response.  An  observation  made  by 
Mead  (18)  in  her  studies  on  the  island  of  Manus  in  New 
Guinea  is  of  interest  in  this  connection.  Dolls  are  ordi- 
narily unknown  to  the  children  on  this  Island.  But  when 
they  were  presented  for  the  first  time  with  some  wooden 
statuettes,  it  was  the  boys  and  not  the  girls  who  accepted 
them  as  dolls,  crooning  lullabies  to  them  and  displaying 
typical  parental  behavior.  This  reaction  is  to  be  under- 
stood in  terms  of  the  pattern  of  adult  behavior  in  Manus. 
Owing  to  the  traditional  division  of  labor,  the  women 
are  busy  with  their  various  duties  throughout  the  day, 
while  the  men  have  much  more  leisure  time  between  their 
activities  of  hunting  and  fishing.  As  a  result  the  father 
rather  than  the  mother  attends  to  the  children  and  plays 
with  them.  This  socially  established  differentiation  of 
behavior  was  reflected  in  the  play  responses  of  the  boys 
and  girls. 

A  vivid  demonstration  of  the  cultural  determination  of 
behavior  is  furnished  by  a  more  recent  study  of  Mead  (19) 
in  which  she  observed  the  traditional  emotional  char- 
acteristics of  men  and  women  in  three  primitive  societies 
in  New  Guinea.  Each  of  these  groups  presented  a  different 
pattern  of  male  and  female  personality.  Among  the 


Arapesh  both  men  and  women  displayed  emotional 
characteristics  which  in  our  society  would  be  labeled 
distinctly  feminine.  In  this  group  both  sexes  are  trained 
to  be  cooperative,  unaggressive,  gentle,  non-competitive, 
and  responsive  to  the  needs  of  others.  They  are  strongly 
imbued  with  a  sense  of  obligation  towards  any  who  are 
weaker  or  younger  than  themselves.  Even  their  typical 
response  towards  material  objects  is  not  one  of  possession 
but  of  solicitude. 

The  MundugumuT)  a  river-dwelling  tribe  of  cannibals 
and  head-hunters,  present  a  sharply  contrasting  picture. 
In  this  society  both  men  and  women  are  violent,  aggres- 
sive, ruthless,  and  competitive.  They  take  great  delight  in 
action  and  in  fighting.  They  are  quick  to  perceive  an 
insult  and  ever  ready  to  avenge  it.  Because  of  an  intricate 
system  of  family  organization,  the  child  is  born  into  a 
hostile  world,  in  which  most  members  of  his  own  sex  are  his 
enemies.  This  is  particularly  true  of  boys,  but  a  child  of 
either  sex  will  be  disliked  and  resented  by  some  members 
of  the  family. 

Perhaps  the  most  interesting  pattern  is  presented  by  the 
Tchambuli)  among  whom  there  is  a  genuine  reversal  of 
the  sex-attitude  of  our  culture.  It  is  the  women  who  have 
the  position  of  power  in  Tchambuli.  The  group  depends 
for  its  food  supply  upon  the  fishing  of  the  women,  the  men 
rarely  engaging  in  this  activity.  Fish  is  also  the  staple 
product  of  trade,  in  exchange  for  which  several  essential 
commodities  are  obtained.  Similarly,  it  is  the  women  who 
make  mosquito  bags,  the  most  important  article  of 
Tchambuli  manufacture  and  in  great  demand  by  outside 
purchasers.  The  men,  on  the  other  hand,  engage  predomi- 
nantly in  artistic  and  non-utilitarian  pursuits.  Most  men 
are  highly  skilled  in  more  than  one  art,  including  dancing, 
carving,  painting,  and  others.  It  is  the  man  in  this  society 


who  is  concerned  with  the  beauty  and  elaboration  of  his 
costumes  and  the  excellence  of  his  artistic  accomplish- 
ments. This  type  of  life  is  reflected  in  pronounced  per- 
sonality differences  between  the  sexes.  The  women  are 
impersonal,  practical,  and  efficient.  Their  attitude  towards 
the  men  is  one  of  kindly  tolerance  and  appreciation.  The 
men  are  graceful,  artistic,  emotionally  subservient,  timid, 
sensitive  to  the  opinions  of  others,  and  throughout  their 
lives  dependent  upon  the  security  afforded  to  them  by  the 

Each  of  these  three  cultures  has  its  "deviants,33  its 
maladjusted  individuals  whose  personality  traits  clash 
with  the  accepted  standards,  as  in  our  society.  But  the 
deviant  in  one  society  often  coincides  with  the  traditional 
ideal  of  another.  Thus  the  "masculine"  woman  among 
the  Tchambuli  is  one  who  embodies  the  typically  feminine 
characteristics  of  our  society;  the  "effeminate"  Tchambuli 
man  displays  behavior  which  we  should  characterize  as 
typically  masculine.  These  observations  inevitably  sug- 
gest an  environmental  interpretation.  In  a  final  evaluation 
of  her  findings,  Mead  writes : 

We  are  forced  to  conclude  that  human  nature  is  almost 
unbelievably  malleable,  responding  accurately  and  contrast- 
ingly to  contrasting  cultural  conditions.  The  differences 
between  individuals  who  are  members  of  different  cultures, 
like  the  differences  between  individuals  within  a  culture,  are 
almost  entirely  to  be  laid  to  differences  in  conditioning,  es- 
pecially during  early  childhood,  and  the  form  of  this  condi- 
tioning is  culturally  determined.  Standardized  personality 
differences  between  the  sexes  are  of  this  order,  cultural  cre- 
ations to  which  each  generation,  male  and  female,  is  trained 
to  conform  (19,  pp.  280-281). 



1.  Baldwin,  B.  T.    "A  Measuring  Scale  for  Physical  Growth 
and   Physiological  Age,"   Nat.   Soc.   Stud.   Educ.,  Fifteenth 
Yearbook,  1916,  Part  I,  11-22. 

2.  .     "The  Physical  Growth  of  Children  from  Birth  to 

Maturity,"  Univ.  Iowa  Stud.  Child  Welfare,  1921,  I.  Pp.  411. 

3.  Book,  W.  F.,  and  Meadows,  J.  L.    "Sex  Differences  in  5925 
High  School  Seniors  in  Ten  Psychological  Tests,"  /.  AppL 
PsychoL,  1928,  12,  56-81. 

4.  Castle,  C.  S.     "A  Statistical  Study  of  Eminent  Women," 
Arch.  Psychol.,  No.  27,  1913.   Pp.  90. 

5.  Cattell,  J.  McK.    "A  Statistical  Study  of  Eminent  Men," 
Pop.  Sci.  Mon.,  1903,  62,  359-377. 

6.  Cattell,  J.  McK.,  and  Cattell,  J.    American  Men  of  Science: 
a  Biographical  Directory.     (Fifth  edition.)    N.  Y.:  Science 
Press,  1933.  Pp.  1278. 

7.  Crampton,  C.  W.    "Anatomical  or  Physiological  Age  versus 
Chronological  Age,"  Fed.  Sem.,  1908,  15,  230-232. 

8.  Ellis,  H.    Man  and  Woman:  a  Study  of  Human  Secondary 
Sexual  Characters.   N.  Y.:  Scribners,  1904.   Pp.  488. 

9.  .  A  Study  of  British  Genius.   London :  Hurst  and  Blac- 

kett,  1904.  Pp.  300. 

10.  Garrett,   H.    E.      Statistics   in   Psychology    and  Education. 
N.  Y.:  Longmans  Green,  1932.   Pp.  317. 

11.  Goodsell,  W.    The  Education  of  Women.    N.  Y.:  Macmillan, 
1923.    Pp.  378. 

12.  Hellman,  M.  "The  Process  of  Dentition  and  Its  Effect  upon 
Occlusion,"  Dental  Cosmos,  1923,  65,  1329-1344. 

13.  Henmon,  V.  A.  C.,  and  Livingstone,  W.  F.    "Comparative 
Variability  at  Different  Ages,"  /.  Educ.  PsychoL,  1922,  13, 

14.  Hollingworth,  L.  S.     "Differential  Action  upon  the  Sexes 
of  Forces  Which  Tend  to  Segregate  the  Feebleminded," 
J.  Abn.  PsychoL,  1922,  17,  35-57- 

15.  Hollingworth,  L.  S.3  and  Montague,  H.    "The  Comparative 
Variability  of  the  Sexes  at  Birth,"  Amer.  J.  Social,  1914-15, 
20,  335-3/0- 


1 6.  Lehman,  H.  C.,  and  Witty,  P.  A.    "Sex  Differences:  Some 
Sources  of  Confusion  and  Error,"  Amer.  /.  PsychoL,  1930, 
42,  140-147. 

17.  Lincoln,  E.  A.    Sex  Differences  in  the  Growth  of  American 
Schoolchildren.    Baltimore:  Warwick  and  York,  1927.    Pp. 

1 8.  Mead,  M.    Growing  up  in  New  Guinea.    N.  Y. :  Morrow, 
1930.   Pp.  372. 

19.  .     Sex  and  Temperament  in  Three  Primitive  Societies. 

N.  Y.:  Morrow,  1935.  Pp.  335. 

20.  Pearson,  K.     The  Chances  of  Death  and  Other  Studies  in 
Evolution.  London: Arnold,  1897.  Vol.1.  Pp.388.  Ch.VIII: 
"Variation  in  Man  and  Woman/' 

21.  Prescott,  D.  A.   "The  Determination  of  Anatomical  Age  in 
School  Children  and  Its  Relation  to  Mental  Development," 
Harvard  Mon.  in  Educ.,  Series  I,  No.  5,  1923.  Pp.  59. 

22.  Pressey,  L.  W.     "Sex  Differences  Shown  by  2544  School 
Children  on  a  Group   Scale  of  Intelligence,  with  Special 
Reference  to  Variability,"  /.  Appl.  PsychoL,  1918,  2,  323- 

23.  Schiller,  B.     "Verbal,  Numerical,  and  Spatial  Abilities  of 
Young  Children,"  Arch.  PsychoL,  No.  161,  1934.   Pp.  69, 

24.  Thorndike,  E.  L.   Educational  Psychology.  N.  Y.:  Teachers 
College,  Columbia  Univ.,  1914.  Vol.  III.  Pp.  408. 


Since  the  beginning  of  the  mental  testing  movement, 
there  has  gradually  accumulated  a  large  body  of  data  on 
sex  differences.  Every  type  of  ability  has  been  covered, 
from  sensory  and  motor  traits,  through  simple  perceptual 
and  associative  tasks,  to  more  complex  intellectual  proc- 
esses and  personality  traits.  Almost  all  tests,  shortly 
after  their  construction,  have  been  administered  to  mem- 
bers of  the  two  sexes,  and  their  scores  compared.  It  was 
a  relatively  easy  task  to  collect  such  data,  especially 
after  the  advent  of  standardized  group  tests.  But  it 
was  quite  a  different  matter  to  determine  what  these 
data  meant.  Itjs  now  apparent  that  the  sex  differences 

^"•^^^MvaawaSws- »W^iwJ»u^« J  .  „,  ,,,,,e          ,    ,„,,,    ,    ,,,,,,.,„  .,,•.•«»•    '  "'  *•*"*""••»  •"'"   ••» 

foujgydjwi^  grottp  reflect  the  social 

jraditions  of  that  group  rather  than  the  innate  .behavioral 
attributes  of  each  sex. 

The  data  which  will  be  surveyed  in  the  present  chapter 
show  sex  differences  under  existing  conditions  in  our  society. 
Such  data,  although  limited  in  their  application,  are  not 
without  value.  Thus  it  is  of  considerable  practical  inter- 
est to  ascertain  the  typical  behavioral  characteristics  of 
men  and  women,  whatever  their  origin.  The  number  of 
situations  in  which  such  knowledge  would  prove  useful 
is  legion.  In  many  fields  of  activity,  definite  assumptions 
are  being  made  in  regard  to  existing  sex  differences  in 
interest,  emotional  appeal,  and  similar  traits.  This  sex 
differentiation  is  noticeable  in  advertising  and  selling, 
political  campaigning,  the  organization  of  newspapers  and 
magazines,  social  work,  crime  prevention,  and  the  treat- 



ment  of  offenders,  to  name  only  a  few  outstanding  exam- 
ples. Insofar  as  cultural  conditioning  may  have  produced 
certain  clear-cut  sex  differences,  these  cannot  be  ignored 
in  the  practical  adjustments  of  everyday  life.  In  a  descrip- 
tive account  of  any  one  cultural  group,  the  question  of 
sex  differences  in  behavior  can  be  legitimately  raised. 

It  is  also  possible  that  a  careful  analysis  of  the  material 
on  sex  differences,  in  conjunction  with  other  available 
information,  may  throw  further  light  upon  the  nature 
and  genesis  of  such  differences.  Such  an  approach  .can 
never  furnish  a  conclusive  account  of  the  origin  of  sex 
differences,  but  it  may  indirectly  yield  some  corroborative 
evidence  on  this  problem. 

In  surveying  the  vast  array  of  studies  on  sex  differences 
in  each  type  of  behavior,  the  general  trend  of  the  results 
will  be  reported.  In  addition,  a  few  of  the  best  illustrative 
studies  in  each  fielcl  will  be  cited.  No  attempt  has  been 
made  to  present  a  complete  survey  of  investigations  on 
this  topic,  since  such  material  has  been  frequently  and 
extensively  reviewed  by  various  writers.1 

In  view  of  the  problems  discussed  in  the  preceding 
chapter,  such  as  selective  factors,  extensive  overlapping 
of  groups,  errors  of  sampling,  errors  of  measurement 
arising  from  inadequacy  of  the  tests,  and  unwarranted 
generalizations  regarding  the  functions  measured,  it  would 
seem  very  difficult  to  formulate  any  summary  statements 
regarding  sex  differences  from  the  data  of  a  number  of 
independent  investigations.  This  is  especially  true  since 
such  investigations  differ  widely  in  number  and  kind  of 
subjects,  specific  tests  or  materials  employed,  and  other 
important  conditions.  Similarly,  all  but  the  most  recent 

1  Cf.,  e.g.,  Allen  (i),  Goodenough  (12),  L.  S.  Hollmgworth  (18,  19),  Lincoln 
(26),  Loutitt  (27),  Wellman  (52),  Woolley  (57,  58),  and  the  more  recent  and 
very  comprehensive  review  by  Miles  (31). 


and  best  controlled  studies  fail  to  report  reliabilities  of 
differences,  degree  of  overlapping,  and  other  essential 
facts,  thus  making  it  difficult  to  evaluate  their  findings. 
In  the  face  of  these  conditions,  the  only  available  criterion 
for  the  acceptance  of  a  conclusion  is  the  consistency  of 
results  of  different  investigators.  A  survey  of  the  experi- 
mental literature  on  sex  differences  reveals  certain  major 
findings  which  are  so  frequently  reported  by  different 
investigators  as  to  suggest  a  valid  basis  in  fact.  It  is 
with  these  findings  that  we  shall  be  primarily  concerned. 


Sex  differences  in  sensory  capacities  are  very  slight, 
ap<;pgQfln^  few.  In  color  discrimination, 

most  investigators  find  women  slightly  superior  (49,  53), 
a  fact  which  may  be  related  to  the  more  frequent  practice 
which  women  have  had  in  the  use  of  color,  as  in  dress, 
household  decorating,  and  embroidery.  Color  blindness  is 
also  less  common  in  the  female,  having  been  observed  in 
about  4%  of  men  and  only  0.5%  of  women  (53).  In 
hearing,  there  seem  to  be  no  significant  sex  differences. 
Although  several  early  investigators  reported  female  su- 
periority in  pitch  discrimination,  more  accurate  experi- 
ments by  Seashore  (41)  have  demonstrated  the  absence 
of  any  such  difference.  In  taste  and  smell,  the  data 
are  too  conflicting  to  allow  any  summary  statement  of 
sex  differences. 

In  the  discrimination  of  lifted  weights,  men  have  gener- 
ally been  found  to  excel  slightly  (53),  This  difference 
has  sometimes  been  attributed  to  the  better  development 
of  motor  ability  in  general  among  men  (cf.  49).  On 
standard  tests  of  sensitivity  to  pain,  women  obtain  lower 
average  thresholds.  Thus  in  experiments  in  which  pain 
was  induced  by  a  steadily  increasing  mechanical  pressure 


on  the  skin,  women  report  a  sensation  of  pain  on  the  aver- 
age sooner  than  men.  It  is  interesting  to  note  that  in 
one  investigation  (n)  this  difference  was  found  to  increase 
with  age,  until  at  the  age  of  1 8  or  19  it  became  over  one 
kilogram.  In  experiments  on  college  students  (56),  the 
average  pressure  at  which  pain  was  reported  was  5.9  kg. 
for  men  and  2.4  kg.  for  women.  It  should  be  noted,  of 
course,  that  such  tests  rely  upon  the  subject's  report  of 
pain.  The  results  are  therefore  open  to  the  influence  of 
social  tradition  and  habits  of  endurance.  In  two-point 
threshold,  i.e.,  the  shortest  perceptible  distance  between 
two  points  on  the  skin,  women  have  also  been  found  to 
have  a  finer  discrimination  (53).  In  the  discrimination 
of  temperature,  degree  of  pressure  on  the  skin,  and  other 
aspects  of  tactual  sensitivity,  no  clear-cut  sex  difference 
has  been  demonstrated  (53). 

In  regard  to  motor  tasks,  some  reviewers  (cf.,  e.g.,  58) 
have  ventured  the  conclusion  that  boys  excel  in  speed 
and  precision  of  movement  in  tasks  in  which  the  direction 
of  attention  remains  fixed,  whereas  girls  excel  in  tasks 
requiring  rapid  adaptations  of  response  and  shifts  of 
attention.  This  generalization,  although  probably  correct 
in  most  cases,  is  not  without  exception.  In  speed  of  tap- 
ping, most  experimenters  report  male  superiority.  Thus 
Thompson  (49)  found  88%  of  the  men  reaching  or  exceed- 
ing the  median  tapping  rate  of  the  women.  Burt  and 
Moore  (6)  reported  that  69.8%  of  the  boys  reached  or 
exceeded  the  girls*  median.  But  a  few  investigators  have 
found  a  slightly  faster  tapping  rate  among  women,  prob- 
ably owing  to  affective  attitudes  and  motivational  factors 
introduced  by  the  conditions  of  the  experiment  (cf.  53, 
I,  p.  140).  Men  have  been  found  to  have  a  shorter  and 
somewhat  more  consistent  reaction  time  than  women. 
In  coordination  tests,  such  as  aiming  and  tracing,  boys 


generally  excel.  Two  functions  in  which  a  very  consistent 
female  superiority  is  reported  are  mirror  drawing  and 
cancellation,  both  involving  rapid  shifts  of  attention  and 
adaptation  to  a  new  mode  of  response  (53). 

The  relation  of  these  findings .  to  the  traditional  play 
activities  of  the  two  sexes  is  apparent.  Boys'  games  offer 
frequent  opportunities  for  the  development  of  rapid  and 
well-coordinated  movements*  Finer  adjustments  of  more 
restricted  muscle  groups,  on  the  other  hand,  are  more  often 
involved  in  the  activities  of  girls. 


A^ljgh^but  highly  consistent  male  superiority  on  tests 
of  mechanical  or  spatial  ^aptitudes  has,  been,  establish^- 
Iri^Thompson's  pioneer  investigation  (49)  on  adult  men 
and  women,  two  mechanical  problems  were  included  in 
the  series  of  tests.  In  both  of  these,  the  subject  was  to 
determine  the  method  of  operation  of  rather  complex 
mechanical  contrivances.  The  average  amount  of  time 
required  for  the  solution  of  each  problem  by  the  male 
subjects  was  considerably  shorter  than  that  required  by 
the  female  subjects.  Similar  results  have  been  obtained 
in  practically  all  subsequent  studies  with  spatial  tests. 
In  the  solution  of  puzzle-boxes,  as  well  as  in  a  variety 
of  form-board  tests,  in  which  the  subject  is  to  insert 
blocks  of  different  shapes  into  the  correct  recesses,  boys 
required  less  time  on  the  average  and  had  fewer  failures 
within  the  allotted  time  limit  than  girls.  In  performance 
on  a  slot  maze,  boys  were  also  found  to  excel  markedly 
(cf.  12). 


&X  iQwXSp^ 

with  girls  of  the  same  Stanford-Binet  I.Q.  Porteus 
regards  this  finding  as  indicative '"  oFTTex"  3itference  in 


planning  capacity,  foresight,  and  similar  personality  traits 
which  he  believes  to  be  measured  by  his  maze  tests.  It 
seems  more  probable,  however,  that  the  obtained  differ- 
ence on  the  two  tests  is  owing  to  the  boys'  superiority  in 
spatial  ability  and  the  girls'  superiority  in  verbal  ability, 
the  latter  playing  a  large  part  in  performance  on  the 
Stanford-Binet.  Similar  results  were  obtained  by  Goode- 
nough  (12)  who  administered  the  Wallin  Peg  Board 
test  to  100  children  between  the  ages  of  2  and  4.  In  this 
test  the  subject  is  required  to  insert  variously  shaped 
pegs  into  the  appropriate  holes  as  rapidly  as  possible. 
A  consistent  sex  difference  was  found  in  favor  of  the 
boys,  although  the  same  group  yielded  a  reliable  difference 
in  favor  of  the  girls  in  Kuhlman-Binet  I.Q. 

In  a  study  (30)  designed  to  measure  "  practical  ability," 
a  series  of  ten  tests  was  administered  to  172  boys  and 
184  girls  between  the  ages  of  10  and  II,  all  of  whom  were 
attending  English  schools.  The  tests  included  four  con- 
struction tests  —  a  wheelbarrow,  a  cradle,  a  girls'  dress, 
and  a  boys'  coat—  as  well  as  a  puzzle  box,  a  "painted 
cube"  test  which  involved  the  ability  to  visualize  spatial 
relations,  and  a  "plunger"  test  measuring  speed  and  co- 
ordination of  movement.  In  all  but  the  two  garment 
teat^the,  ,,bpy  s.  pb  taiued  higher  scores.  It  was  also  found 
that  the  boys'  scores  on  the  different  spatial  tests  corre- 
lated more  highly  with  each  other  and  less  highly  with 
estimates  of  "general  intelligence"  than  did  the  girls' 
scores.  Frg>m  this  it  was  suggested  that  performance  on 
such  tests  depends  largely  upon 
s*  it  is 

bys  those  processes  which  are  included  under  the  heading 
of  "general  intelligence."  Tire  method  of  arriving  at  a 
solution  of  the  same  problem  might  thus  have  differed 
for  the  two  sexes. 


In  another  investigation  (10)  on  English  schoolchil- 
dren, 52  boys  and  48  girls  were  given  a  series  of  tests  se- 
lected from  various  current  performance  and  non-language 
scales.1  When  scores  on  all  the  tests  were  lumped  into 
a  single  measure  and  I.Q.'s  computed  from  it,  no  signifi- 
cant sex  difference  was  found,  the  average  I.Q.  of  the 
boys  being  96  and  that  of  the  girls  93.  Analysis  of  per- 
formance on  separate  tests,  however,  revealed  several 
marked  sex  differences.  In  ^general  it  was  concluded 
that  girls  excel  in  tests  of  memory,  learning,  and  non- 
form  relationships,  and  boys  excel  in  tests  of 

geometrical  form  relationships  and  reasoning. 

Stenquist  (43)  reports  that  on  his  mechanical  aptitude 
tests,  elementary  school  girls  obtained  only  about  65% 
as  many  correct  items  as  boys,  and  that  women  in  grad- 
uate school  obtained  about  80%  as  many  correct  items 
as  unselected  adult  m,pn.  In  ,an  extensive  investigation 
conducted,  with,  the  Minnesota  Mechanical  Aptitude  Tests 
(37),  similar  differences  were  found.  Seventh  grade  boys 
anXgirls,  as  well  as  male  and  female  college  sophomores, 
were  employed  in  the  sex  comparisons.  In  Table  XVII 
are  shown  the  critical  ratios  (s^.tfaffference)  °f  ^e  differences 
between  male  and  female  averages  in  each  of  these  two 
populations.  Sex  differences  were  found  in  single  tests 
as  well  as  in  Batteries  A,  B,  C,  and  D.  These  bat- 
teries were  composites  of  several  tests,  each  test  being 
weighted  so  as  to  yield  the  best  possible  estimate  of 
mechanical  aptitude. 

Those  tests  in  which  male  superiority  was  most  marked 
involved  the  use  of  mechanical  materials  which  would 
be  more  common  in  the  environment  of  boys.  In  block 
packing  and  card  sorting,  female  superiority  is  usually 

1  Including  Pintner-Paterson,  Army  Performance  Scale,  and  Porteus  maze 






(After  Paterson,  et  at.,  37,  p.  274) 


Critical  Ratios  * 

Seventh  Grade  Pupils 

College  Sophomores 

Minnesota  Assembly 



Battery  B 



Battery  A 



Battery  C 



Paper  Form  Board 



Battery  D 



Minnesota  Spatial  Relations 



Packing  Blocks 



Card  Sorting 



*  In  this  table  a  minus  sign  indicates  a  difference  in  favor  of  the  girls. 

found.  These  tests  involve  chiefly  visuo-motor  coordina- 
tion with  rapid  shifts  of  attention,  in  which  women 
generally  excel  (cf.  above). 

In  numerical  aptitude  a  t  .tendency  towards  male  su- 
periority has  likewise  been  found,  although^are 

ao.t  so  consistent,  as, m  Jt&fi ...£&$$  pf J^echanical  aptkude. 

This  also  is  to  be  expected  from  social  tradition  and  en- 
vironmental conditions,  since  the  sex  differentiation  in 
regard  to  mathematical  work  is  not  so  pronounced  as  it  is 
for  mechanical  pursuits.  Girls  are  afforded  relatively  more 
opportunity  for  the  development  of  mathematical  abilities 
than  for  the  exercise  of  mechanical  functions.  In  the  ele- 
mentary school,  for  example,  girls  are  taught  arithmetic 
in  the  same  classes  as  boys,  but  the  latter  are  still  segre- 
grated  for  "shop  courses." 

A  few  studies  on  the  development  of  number  concepts 
in  p  re-school  children  yield  conflicting  data  (cf.  12).  These 
studies  are  difficult  to  interpret  on  account  of  the  relatively 


crude  techniques  employed,  and  also  because  of  the  close 
linkage  between  number  concepts  and  language  develop- 
ment in  the  young  child.  Test  results  on  older  subjects 
indicate  a  slight  but  fairly  general  male  superiority  in 
both,  the  mechanics  of  computation  and  mathematical 

An  analysis  of  the  data  employed  in  the  standardization 
of  the  Stanford-Binet  showed  that  the  average  percentage 
of  boys  who  succeeded  on  the  numerical  tests  was  greater 
than  that  of  girls.  The  difference  in  favor  of  the  boys  was 
33%  on  the  arithmetic  reasoning  test,  15%  on  the  test 
of  making  change,  and  10%  on  the  "induction"  test  in 
which  a  generalized  numerical  rule  must  be  discovered 
(46,  p.  82).  In  an  investigation  on  German  schoolchildren 
(8),  a  continuous  addition  test  was  administered  to  1214 
boys  and  girls  in  grades  i  to  8.  In  almost  every  grade, 
the  boys  completed  more  additions  within  the  time  allowed 
and  made  fewer  errors  than  the  girls.  An  extensive  in- 
vestigation on  various  aspects  of  mathematical  aptitude 
was  conducted  on  500  boys  and  girls  in  English  secondary 
schools.  The  tests  covered  the  following  fields:  arithmetic 
computation,  arithmetic  reasoning,  mechanical  algebraic 
processes,  the  use  of  algebraic  symbols  in  problem  solving, 
the  manipulation  of  geometric  constructs  and  spatial  rela- 
tionships, and  geometric  reasoning.  It  will  be  noted  that 
both  mechanical  and  numerical  abilities  are  involved  in 
such  tests.  A  small  but  consistent  difference  in  favor  of 
the  boys  was  found  on  the  series  as  a  whole,  larger  and 
more  significant  differences  being  obtained  on  certain 
specific  tests. 

Since  1930  a  mathematical  section  has  been  added  to 
the  Scholastic  Aptitude  Tests  administered  by  the  College 
Entrance  Examination  Board  to  prospective  college  stu- 
dents (35).  A  large  and  significant  sex  difference  in  per- 


formance  on  this  .part  of  the  test  has  been  found,  as  is 
illustrated  by  the  1930  results  reported  below. 

^  .         Critical  Ratio 
Boys  Girls      of  Dyffrence 

Number  of  cases  4394  33  18 

Average  score  in  mathematical 

section  of  Scholastic  Aptitude 

Test  510-22        484-37  "-59 

It  will  be  seen  that  the  difference  between  the  male  and 
female  averages  is  11.59  times  as  l^arge  as  its  standard 
error,  which  is  far  in  excess  of  the  minimum  critical  ratio 
of  3  required  for  perfect  reliability.  Similar  results  have 
been  obtained  in  subsequent  years.  Itis  interesting  to  note 
that  this  large  and  highly  reliable  difference  '  was  Build 
among  subjects  who  had  been  exposed  to  the,  differentiat- 
ing effects  of  environment  over  a  considerable  period 
of  time,  and  that  the  corresponding  differences  among 
younger  subjects  were  far  less  pronounced. 

In  verbal  or  linguistic  aptitude  there  is  a  fairly  consistent 
difference  "m  '""favor  "oTltTTemale''  sex.  These  differences 
"are  manifested  frbitr  '"a""  Very  early"  age.1  Observations  on 
normaraTwell  as  ""gjftecf  an'd  feebleminded  children  have 
shown  that  on  the  average  girls  begin  to  talk  earlier  than 

boys.    SroUsui^  a£e  ka^e^,!fr£er  vo- 

cabulary than  boys.  In  .one^^vestigation  on  51  children 
at  the  age  of  18  months,  the  average  number  of  words 
employed  by  the  girls  was  found  to  be  78.6  -and  by  the 
jDoys  only  59.  In  another  study  the  percentage  of  compre- 
hensible verbal  responses  was  determined  for  each  child. 
The  average  percentages  proved  to  be  14  and  49  for  boys 
at  1  8  and  24  months,  respectively,  as  compared  to  37  and 
78  for  girls  at  the  same  ages.  Speech  defects  ..and  leading 

1  Cf.  McCarthy  (29)  for  a  survey  of  sex  differences  in  early  language  develop- 



disabilities  are  also  less  common  among  girls  than  among 
boys.  Similar  results  have  been  obtained  with  all  other 
indices  of  linguistic  development  (cf.  29,  52). 

Female  superiority  on  linguistic  tests  persists  through 
the  successive  academic  levels.  It  is  equally  apparent 
among  elementary  schoolchildren  and  college  students. 
Female  superiority  has  been  consistently  reported  in 
speed"  of  reading,  naming  opposites,  and  in  all  verbal 
association  tests  (53).  Girls  excel  on  sentence  completion 
tests  (12,  53)5  and  a  similar  difference  has  been  found  in 
tests  of  story  completion  (53).  On  the  verbal  part  of  the 
Scholastic  Aptitude  Test,  a  large  and  highly  reliable  sex 
difference  has  been  obtained  in  the  reverse  direction  from 
that  found  in  the  numerical  part  (35).  The  average  scores 
of  the  two  sexes,  as  well  as  the  number  of  cases  in  each 
group  and  the  reliability  of  the  difference,  are  shown1 

0  « .  ,          Critical  Ratio 

Boys  Girls        of  Differgnce 

Number  of  cases  4394  3318 

Average  score  in  verbal  section 

of  Scholastic  Aptitude  Test      486.98         5*7'54  X3'52 

It  is  particularly  significant  that  this  reversal  in  the  rela- 
,  5ver*stan3nig  of  the  two  sexes  was  obtained  within  the 
same  group  of  subjects.  It  cannot  therefore  be  attributed  to 
selective  'factors  or  other  spurious  influences. 

Finally  we  may  consider  the  results  obtained  in  a  large 
number  of  studies  on  memory.    On  practically  all  tests  of 

*ur**. .*¥**•'  6***fc^^^ 

memory,  a  sex  difference  in  favor  of  t$g ,1;^^].^ has-been 
found.  As  in  the  case  of  verbal  ability,  these  differences 
have  been  observed  very  early  in  life  and  persist  at  all 
ages.  There  is  considerable  evidence  to  show  female 
superiority  in  memory  among  pre-school  children.  Terman 
(45),  in  a  study  on  112  kindergarten  children  with  the 


Stanford-Binet,  found  the  girls  superior  on  the  tests  of 
digit  span  and  memory  for  sentences.  Among  the  results 
obtained  with  the  Stanford-Binet  on  older  children,  Ter- 
man  reports  girls  superior  in  memory  for  digits  and  repro- 
duction of  drawings  from  memory  (46). 

In  a  number  of  studies  (cf.,  e.g.,  34,  40),  girls  have  been 
found  to  excel  more  markedly  in  logical  than  in  rote  mem- 
ory. This  is  doubtlessly  owing  to  the  greater  dependence 
of  logical  memory  tests  upon  verbal  comprehension.  In  a 
few  cases  in  which  the  specific  material  of  the  test  was 
more  familiar  or  more  interesting  to  boys,  the  relative 
standing  of  the  two  sexes  was  reversed.  But  these  in- 
stances are  very  infrequent,  and  on  the  large  majority  of 
memory  tests  female  superiority  is  found.  This  superiority 
has  been  established  for  a  wide  range  of  materials,  for 
recall  as  well  as  recognition,  and  direct  as  well  as  incidental 
memory.  In  this  connection  mention  may  also  be  made 
of  the  fact  that  women  have  more  vivid  mental  imagery 
than  men  in  every  sense  modality.  This  finding,  first  sug- 
gested by  Galton  on  the  basis  of  his  famous  "breakfast 
table"  questionnaire,  has  been  subsequently  corroborated 
by  several  investigators. 

Further  data  on  sex  differences  in  these  various  apti- 
tudes will  be  discussed  in  the  following  sections,  in  con- 
nection with  intelligence  test  performance  and  records  of 
scholastic  achievement.  The  material  on  verbal  and  nu- 
merical aptitudes,  in  particular,  needs  to  be  largely  sup- 
plemented from  intelligence  test  results.  Since  these 
abilities  have  not  been  generally  recognized  as  distinct 
from  "general  intelligence,"  there  are  relatively  few  stud- 
ies on  sex  differences  in  verbal  or  numerical  abilities  as  such. 

In  reference  tp  the  possible  origin  of  sex  differences  in 
special  aptitudes,  a  comment  offered  by  Burt  several  years 
ago  is  of  interest.  He  states: 


Sheltered,  supervised,  detained  at  home,  girls  .  .  .  incline 
to  sedentary  lives  and  engage  in  literary  pursuits;  and  .  .  . 
they  consequently  excel  in  linguistic  work  and  conversational 
activities.  Boys  .  .  .  have  more  to  do  with  practical,  per- 
ceptual, out-of-door  pursuits.  They  are  sent  to  shops  with 
money.  They  are  allowed  to  play  and  wander  in  the  streets. 
They  are  encouraged  to  handle  tools — to  construct  toys  for 
amusement  and  articles  for  use.  No  wonder  that  .  .  .  boys 
grow  more  ready  with  hand  and  eye  than  with  tongue  or 
pen  (5,  p.  196). 


On  the  Binet  test  and  its  revisions.,  sex  differences  are 
sligEFBuf  usually  favor  the  girls  up  to  the  age  of  about  14. 
Terman  (44)  reports  results  obtained  with  the  Stanford- 
Binet  on  457  boys  and  448  girls  between  the  ages  of  5  and 
14.  The  average  I.Q.  of  the  girls  was  found  to  be  higher 
than  that  of  the  boys  at  every  age  except  10,  at  which  they 
were  equal,  and  at  14,  at  which  the  boys  excelled.  The 
specific  data  for  each  age  group  are  reproduced  below. 

Age  Boys  Girls 

5  ioo  104 

6  99  105 

7  101  103 

8  IOO  IO2 

9  98  102 

10  103  103 

11  96  101 

12  97  99 

13  96  97 

14  ioo  96 

These  differences,  although  slight,  are  consisferit  .iyith 
the  findings  of  other  investigators,  Burt  (5)  tested  3500 
London  schoolchildren,  ranging  in  age  from  3  to  14,  with 
his  own  revision  of  the  Binet  tests.  With  the  exception  of 
a  single  age  group,  the  girls  obtained  higher  average  mental 


ages  than  the  boys.  In  general  this  difference  amounted  to 
about  -j-  of  a  year.  Goodenough  (13)  administered  the 
Kuhlman-Binet  to  50  boys  and  50  girls  at  each  of  the  ages 
2,- 3,  and  4.  The  average  I.Q.  of  the  girls  was  again  higher 
than  that  of  the  boys,  this  difference  persisting  when  the 
children  were  retested  after  a  lapse  of  six  weeks.  Thus  the 
oBtained  difference  could  not  be  attributed  to  chance 
fluctuations  or  a  response  to  novel  situations. 

With  most  group  tests,  a  similar  female  superiority  up 
to  the  age  of  14  has  been  found.  Thus  in  several  extensive 
school  surveys  with  the  National  Intelligence  Test,  girls 
have  obtained  consistently  higher  scores  than  boys  (cf., 
e.g.,  54).  On  the  Dearborn  group  scale,  which  is  less  de- 
pendent on  verbal  ability  than  most  group  scales,  however,  no 
consistent  sex  difference  was  found  in  a  group  of  3400 
children  between  the  ages  of  7  and  17  (cf.  25). 

Among  high  schqol  .students,  on  the  other  hand,  boys 
generally  excel  on  intelligence  tests.  'Tharndfelso),  using 
a  group  mtelligenc^.^q.ak,  especially ,  .constructed  for  the 
high  school  level,  with  about  2500  high  school  students 
between  the  ages  of  13  and  18,  found  the  boys  to  be 
superior  on  the  average.  Similarly,  in  an  investigation  on 
1453  high  school  students  with  the  South  Dakota  Group 
Intelligence  Test,1  boys  were  found  to  excel,  their  superior- 
ity increasing  from  the  first  to,  the  last, year,  of.  high  school 
(3).  In  a  number  of  studies  employing  the  Army  Alpha 
with  both  high  school  and  college  groups,  consistent  male 
superiority  was  also  found  (cf.,  e.g.,  55). 

These  sex  differences  in  total  scores  on  intelligence  tests 
can  be  explained  in  terms  of  certain  previously  reported 
findings.  The^^en^^  age 

on  the  Binet  tests,  as  well  as  on  most  ofjthe  groi^p  tests, 

3  Including  tests  of  directions,  analogies,  synonym-antonym,  reading  com- 
prehension, information,  and  arithmetic. 


is  to  be  expected  on  the  basis  of  the  large  verbal  content  of 
such  scales  and  their  frequent  use  of  memory  tests.  It  will 
be  recalled  that  girls  excel  in  both  of  these  aptitudes.  This 
interpretation  of  female  superiority  in  general  intelligence 
tests  is  supported  by  the  fact  that  the  sex  difference  dis- 
appears or  becomes  reversed  when  tests  like  the  Dearborn, 
which  lay  less  emphasis  upon  verbal  ability,  are  employed. 
The  consistent  male  superiority  on  Army  Alpha  has  like- 
wise been  attributed  to  the  fact  that  this  scale  was  orig- 
inally standardized  on  men  and  thus  may  contain  an  ex- 
cess of  items  which  favor  men.  FiagHy,, the  reversal  9$  ^e 
relative  status  of  the  two  sexes  in  high  school  groups  has 
already  been  explained  in  terms  of  selective  factors  (cf. 
Ch.  XIV).  Even  when  predominantly  verbal  tests  are 
employed,  male  superiority  is  found  among  high  school 
students.  The  fact  that  this  superiority  of  the  boysftends 
to  increase  throughout  the  four  years  of  high  school  fur- 
ther corroborates  the  proposed,  explanation. 

That  sex  differences  on  tests  of  so-called  general  intelli- 
gence are  attributable  to  differences  in  special  aptitudes  is 
also  indicated  by  an  analysis  of  scores  on  the  separate 
subtests  of  intelligence  scales.  Thus  an  examination  of  the 
scores  obtained  by  834  high  school  students  on  Army 
Alpha  showed  male  superiority  only  on  the  arithmetic 
reasoning,  number  series  completion,  and  information 
tests  (55).  The  differences  on  these  tests  were  sufficient  to 
pull  up  the  total  scores  and  produce  a  sex  difference  in 
favor  of  the  boys  on  the  scale  as  a  whole. 

In  a  previously  mentioned  investigation  with  the  Pres- 
sey  Group  Test  (cf.  Ch.  XIV),  the  scores  of  880  children 
between  the  ages  of  9  and  14  were  examined  with  reference 
to  performance  on  each  of  the  ten  subtests  of  the  scale. 
It  will  be  recalled  that  at  these  ages  the  girls  excelled  in 
total  score.  On  the  separate  tests,  however,  the  boys 


excelled  in  arithmetic  at  all  ages  and  in  practical  informa- 
tion from  age  n  on.  The  girls  excelled  in  rote  memory  for 
words,  naming  opposites,  word  completion,  dissected  sen- 
tences, analogies,  and  moral  classification,  the  last  depend- 
ing largely  upon  verbal  comprehension.  In  the  remaining 
two  tests,  no  significant  sex  difference  was  exhibited  (39). 
It  will  be  recalled  that  among  high  school  seniors,  owing 
to  a  differential  elimination  of  male  and  female  students, 
boys  excelled  in  total  scores  on  the  Pressey  test  (4).  Per- 
formance on  the  separate  tests,  however,  showed  the  same 
hierarchy  as  in  the  younger  groups.  Thus  the  boys  ex- 
celled markedly  in  the  arithmetic  and  information  tests. 
The  girls  obtained  higher  average  scores  in  word  com- 
pletion, dissected  sentences,  and  logical  memory.  On  the 
remaining  tests  there  was  either  no  difference  or  a  slight 
superiority  of  the  boys.  The  reversal  in  total  score  from 
the  elementary  to  the  high  school  groups  results  from  the 
fact  that  the  high  school  senior  boys  excelled  by  a  much 
larger  amount  in  those  same  tests  in  which  elementary 
school  boys  excelled.  Similarly,  the  high  school  senior 
girls  excelled  by  a  smaller  amount  in  the  tests  in  which  the 
elementary  school  girls  had  excelled  markedly. 


On  the  whole  gMsjexcsi JjL^£2£!^./£Ag/Arrf^.  achieve- 
ment, as  revealed  botkiy  achievement .test  records aiiyd  by 
school  grades.  Performance  on  the  separate  parts 
ardized  achievement ,  tests  y  however,-  shows ,  a,  JbLi 
abilities  in  different  school  subjects,  which  corresponds 
closely  to  that  found  with  tests  of,  iiitjp,llige:ncewaiid-  special 
aptitudes.  In  an  investigation  (8)  on  300  schoolchildren 
with  the  StanfbKTlKffi^  sex  (difference  was 

"fouiid  in  total  achievement  "quotient   But  an  analysis  of 
the  scores  indicated  a  definite  superiority  of  the,  .girls  in 


reading,  .language. .usage,  and  dictation,  and  of  the  boys 
in  arithmetic,  nature  study,  and  history.  In  another 
investigation  (51)  a  standard  geometry  test  was  admin- 
istered to  410  high  school  boys  and  349  high  school  girls, 
all  of  whom  had  previously  been  given  the  Terman  Group 
Test  of  Intelligence.  The  boys  as  a  group  were  found  to 
excel  in  geometric  ability  when  compared  with  girls  of  equal 
"intelligence"  as  indicated  by  the  Terman  scores. 

Numerous  other  extensive  school  surveys  have  yielded 
the  same  general  results.1  In  general  girls  obtain  higher 
achievement  test  scores  in  subjects  requiring  verbal  abil- 
ity or  memory.  Boys  excel, in  those  subjects  which  call 
into  play  numerical  or  spatial  aptitudes,  as  well  as  in  cer- 
tain "information"  subjects  such  as  history,  geography, 
and  general  science.  This  is  in  agreement  with  the  common 
superiority  of  boys  on  tests  of  general  information,  and 
probably  results  from  the  less  restricted  and  more  hetero- 
geneous environment  to  which  boys  are  exposed  as  well 
as  from  their  wider  range  of  reading  interests.  Thus  Ter- 
man (47),  in  his  survey  of  the  reading  habits  of  gifted 
children,  reports  that  the  girls  read  imaginative  and  emo- 
tional fiction  as  well  as  stories  of  school  and  home  life  far 
more  often  than  the  boys,  while  the  latter  showed  a  pre- 
dominant interest  in  books  on  science,  history,  biography, 
travel,  and  informational  fiction  and  adventure  tales. 

In  regard  to  school  progress,  girls  are  consistently  more 
successful ""tTiaa  Boys.  The  differences,  although  small, 
appear  irrespective  of  the  particular  criterion  of  school 
progress  employed.  Girls  are  less  frequently  retarded, 
more  frequently  accelerated,  and  promoted  in  larger  num- 
bers than  boys.  Typical  results  from  a  survey  conducted 
in  the  schools  of  318  cities  are  shown  in  Table  XVIII. 
Since  girls  make  more  rapid  progress  than  boys  in  school 

1  Cf.  Lincoln  (26),  Ch.  IV,  for  a  survey  of  this  material  previous  to  1927. 



promotions  and  since  most  comparisons  of  achievement 
are  made  on  the  basis  of  school  grade,  it  follows  that  age 
comparisons  would  show  an  even  greater  superiority  of 

In  school  grades  girls  excel  consistently  even  in  those 
subjects  which  favor  boys.  Thus  a  comparison  of  grades  in 
arithmetic,  or  history,  or  any  "other  subject  in  which  boys 
obtaiiTKigKer  achievement  test  scores,  shows  a  sex  differ- 
encelh  favor  of  girls.  The  advantage  enjoyed  by  girls  in 
school  grades  was  made  particularly  vivid  in  an  investiga- 
tion (25)  on  202  boys  and  188  girls  in  grades  2  to  6,  all  of 
whom  were  given  the  Stanford  Achievement  test.  The 
girls  were  found  to  excel  consistently  in  school  grades, 
when  compared  with  their  achievement  test  scores.  Thus  the 
grades  showed  a  far  greater  female  superiority  than  seemed 
to  be  warranted  by  performance  on  objective  achievement 




(After  Lincoln,  26,  p.  100) 

School  Status 

Cities  of  over  25,000 

Cities  of  less  than 










I  year  over  age 





2  years  over  age 





3  years  over  age 





4  years  over  age 





Total  over  age 





Total  under  age 





v.,  high   school   girls   generally   obtain    better 
grades  than  high  school  boys,  even  though  the  latter  are 


a  more  select  group  and  make  a  ^better  $  ho  wing,,  o,a  in- 
telligence tests  (cf.  36).  This  is  illustrated  by, a  survey  of 
the^grades^given  to  students  in  each  of  the  four  years  of  a 
single  high  school,  the  results  of  which  are  shown  in 
Table. XIX.  It  will  be  noted  that  in  each  year,  without  a 
single  exception,  the  percentage  of  A's  and  B's  is  larger 
and  the  percentage  of  D's  and  F's  is  smaller  among  the 
girls  than  among  the  boys.  The  larger  percentage  of  boys 
than  girls  who  left  school  further  suggests  the  better  ad- 
justment of  girls  to  the  school  situation. 




(After  Lincoln,  26,  p.  93) 


First  Year 

Second  Year 

Third  Year 

Fourth  Year 






































21.  1 
















Left  school 









Various  explanations  have  been  offered  for  the  greater 
scholastic  success  of  girls.  Among  the, i(3^ajp.j^faiKti5«wnay 
be  mentioned  girls'  dem9nstra^^ 

altitude,  whi<&  prqbablyplays  an  in^port^xjJjjjL^ij^^jxlj^ 
all  school  subjects.  Current  methods  of  instruction,  as 
well  as  methods  of  testing,  are  predominantly  verbal.  The 
child  who  expresses  himself  well,  furthermore,  will  impress 
the  teacher  as  being  relatively  brighter  than  one  who  is 
linguistically  backward,  and  this  may  in  turn  affect  their 
respective  grades.  A  second  possible  factorjn^jthe^higher 
•lastic  ratings  of  ^girl?TF:Sie^li^tness  and  general  su- 

"—•*--*-—„ _..— £L» ^«^**Q— — —»«- i«**»^J*a«w»i!«-wmt^^,v^w,^,^1.  ,.*. 



periority  of  their  handwriting.  In  most  investigations  on 
both  elementary  and  high  school  groups,  girls  have  been 
found  to  excel  markedly  in  the  quality  of  their  handwrit- 
ing, as  judged  by  standardized  product  scales  (cf.  26, 

pp.  72-77)- 

Finally  owing  to  the  obvious  presence  of  a  subjective 
element  in  school  grades,  personality  differences  between 
boys  and  girls  may  influence  the  allotment  of  such  grades. 
Girls  are  generally  more  docile,  quieter,  less  resistant  to 
school  discipline,  and  are  less  often  "behavior  problems" 
than  boys.  This  difference  in  the  child's  attitude  towards 
school  affects  his  grades  both  through  the  amount  of  ma- 
terial actually  learned  and,  more  directly,  through  the 
general  impression  created  on  the  teacher. 


That  definite  personality  differences  exist  between  adult 
men  and  women  in  our  society  is  clearly  apparent  from 
everyday  observation.  In  many  emotional  and  social 
characteristics,  this  differentiation  is  noticeable  from  an 
early  age.  An  important  aspect  of  personality  develop- 
ment in  which  traditional  sex  differences  are  manifested 
includes  interests,  preferences,  ideals,  attitudes,  and  per- 
sonal sense  of  values.  These  characteristics,  because  of 
their  relatively  subtle  and  persistent  nature,  often  exert 
an  unsuspected  influence,  not  only  upon  the  development 
of  emotional  and  character  traits,  but  also  upon  the  in- 
dividual's achievements  and  effective  abilities.  An  in- 
teresting observation  recorded  by  Thompson  (49)  illus- 
trates this  effect.  In  discussing  the  superior  performance 
of  a  group  of  men  as  contrasted  to  a  group  of  women 
on  a  problem  of  mechanical  ingenuity,  she  wrote: 

It  is  very  difficult  to  evaluate  this  difference  because  of 
the  indefinite  nature  of  the  problem.     Most  of  the  women 


expressed  a  great  distaste  for  all  such  problems,  because  they 
were  uninteresting.  Many  of  them  were  so  uninterested  that 
they  did  not  really  work  at  it.  Whether  the  men  found  it 
equally  uninteresting,  but  forced  themselves  to  work  in  spite 
of  lack  of  interest,  or  whether  the  problem  appealed  to  them 
as  more  interesting,  it  is  difficult  to  say.  From  the  voluntary 
comments  of  the  subjects  the  latter  hypothesis  seems  more 
probable  (49,  p.  116). 

It  has   proved   quite   difficult  to  obtain  objective  or 
quantiB;tfv^M*<S'asttfes  of  these  characteristics.   A  number 

j^  ^  ^^  differences  employed 

the  free  association  technique,  whereby  the  subject  is 
presented  with  a  list  of  stimulus-words  and  is  required  to 
respond  by  giving  the  first  word  suggested  to  him  by  each 
stimulus-word.  Although  admittedly  crude,  these  early 
experiments  yielded  fairly  consistent  results  which  were 
also  in  general  agreement  with  common  observation.1  In 
general  women's  associations  seemed  to  indicate  an  in- 
tefestTTrT  tfie  immediate  surroundings,  finished  products, 
the  ornamental,  the  individual,  and  the  concrete,  whereas 
men's  associations  suggested  an  interest  in  the  more  re- 
mote, constructive,  useful,  general,  and  abstract.  Dynamic 
concepts  were  more  common  among  the  men's  responses, 
static  among  those  of  the  women.  Time  was  more  prom- 
inent in  the  men's  associations,  space  in  the  women's. 
The  free  association  responses  of  women  also  tended  to 
be  more  subjective,  personal  or  particular,  social,  and 
emotionally  toned  than  those  of  men. 

Equally  characteristic  and  traditional  ^differeace^  were 
revealed  in  a  large  number  of  subsequent  studies  on,  qhil- 
dren.  Sex  differences  were  investigated  through  an  analysis 
of  literary  and  reading  preferences,  spontaneous  con- 
structions and  drawing  choices,  aesthetic  preferences, 

1  For  a  survey  of  these  studies,  see  Miles  and  Terman  (32). 


vocational  choices,  and  choice  of  ideals.  Thus  in  one 
typical  investigation  (21),  200  schoolchildren  were  re- 
quired to  rank  nine  pictures,  each  representing  a  specific 
type  of  appeal,  in  order  of  merit.  The  chief  sex  difference 
noted  was  the  relatively  high  position  assigned  by  girls 
to  pictures  dealing  with  children,  and  by  boys  to  those 
whose  main  theme  was  action  and  the  heroic. 

A  series  of  more  recent  studies  have  approached  the 
same  problem  through  an  analysis  of  conversations.  In  the 
first  of  these  investigations  (33),  a  record  was  made  of 
conversations  heard  in  the  evening  along  a  busy  thorough- 
fare in  the  heart  of  New  York  City's  theatre  district. 
When  the  participants  were  classified  according  to  sex, 
it  was  found  that  the  most  frequent  topics  of  conversation 
among  women  were  clothes  and  social  affairs,  and  among 
men,  money  and  business  affairs.  In  conversations  between 
men  and  women,  there  was  a  predominance  of  men's  conver- 
sational topics,  which  suggested  a  tendency  for  women  to 
adapt  themselves  to  the  interests  of  their  male  companions. 

In  a  subsequent  study  (24),  conversations  were  collected 
in  a  wider  variety  of  situations  in  a  mid-western  city. 
Observations  were  made  on  a  college  campus,  in  barber 
shops,  churches,  hotel  lobbies,  street  cars,  and  other  public 
or  semi-public  places.  Although  typical  conversational 
differences  were  noted  from  one  locale  to  another,  the 
general  results  corroborated  those  of  the  New  York  study. 
The  two  major  topics  in  order  of  their  frequency  were, 
among  women,  (i)  men  and  (2)  clothes;  and  among  men, 
(i)  business  and  money,  and  (2)  sports  and  amusements! 
It  is  also  interesting  to  note  that  " people"  of  either  sex 
played  a  larger  part  in  the  conversations  of  women  than 
in  those  of  men,  having  been  observed  in  37%  of  the 
former  and  only  16%  of  the  latter.  A  similar  series  of 
observations  was  made  in  two  London  streets  (23).  The 


results  resembled  closely  those  of  the  American  studies, 
with  two  major  exceptions.  First,  in  all  conversations 
business  and  money  were  less  prominent,  and  secondly, 
in  conversations  between  men  and  women  there  seemed 
to  be  a  greater  tendency  for  the  men  to  adapt  to  women's 
interests  than  in  the  American  groups. 

Qonsistent  sex  differences  have  also  been  reported  in 
th<e  individual's  conception  of  values.    The  application  of 








•  Men  (N=463) 
Women  (N*313) 




Theoretical       Economic        Aesthetic          Social  Political          Religious 

Type  of  Value 

ALLPORT-VERNON  STUDY  OF  VALUES.  (Data  from  Allport  and  Vernon,  2, 
p.  246.) 

the  Allport-Vernon  "Study  of  Values"  (2)  has  furnished 
some  information  on  these  differences.  This  is  a  test  of 
the  questionnaire  type,  in  which  each  response  is  given  a 
numerical  weight  in  one  of  six  sets  of  values,  viz.,  the- 
oretical, economic,  aesthetic,  social,  political,  and  religious. 
The  relative  standing  of  the  subject  in  each  set  is  then 
determined  by  totaling  scores  on  separate  items,  and  a 
psychograph  is  constructed  -from  these  six  measures, 
Psychographs  showing  the  average  scores  of  463  men  and 
313  women  will  be  found  in  Figure  48.  It  will  be  noted 


that  among  the  women  the  highest  average  values  fall 
in  the  esthetic,  social.,  and  religious  groups,  in  the  order 
given,  while  among  the  men  they  fall  in  the  theoretical, 
economic,  and  political. 

In  connection  with  the  higher  rating  of  social  values 
among  women,  mention  may  be  made  of  the  widely  quoted 
female  superiority  in  social  perception.  Relatively  little 
information  is  available  on  this  point,  but  a  comparison 
of  the  performance  of  men  and  women  on  the  George 
Washington  Social  Intelligence  Test  may  prove  relevant. 
In  a  group  of  430  college  students  of  both  sexes  who  took 
this  test,  a  small  difference  in  total  score  was  found  in 
favor  of  the  women  (20).  The  average  scores  obtained  by 
freshmen  and  upperclassmen  of  each  sex  were  as  follows: 

Freshmen  Upperclassmen 

Men  101  109 

Women  108  115 

Upon  analysis,  this  difference  was  found  to  be  owing  al- 
most entirely  to  the  higher  scores  of  the  women  students 
on  two  subtests.  One  of  these  was  designed  to  measure 
the  ability  to  decide  the  "best  thing  to  do"  in  a  difficult 
social  situation.  The  other  involved  a  knowledge  of  com- 
mon facts  of  human  behavior  which  could  be  acquired 
through  the  everyday  observation  of  people*  In  view 
of  the  larger  part  played  by  women  in  social  situations 
and  the  more  frequent  concern  of  women  with  the  reac- 
tions of  others,  such  differences  are  easily  understand- 


Those  aspects  of  personality  commonly  .designated  as 
moral  of  character  traits  also  present  Atypical ^differences 
between  the  sexes.  In  an  investigation  (42)  on  tnoraT 


judgment,  a  list  of  100  acts,  including  commendable  as 
well  as  reprehensible  activities  all  of  which  might  be  car- 
ried out  by  a  young  man  of  18,  were  to  be  rated  on  a  scale 
ranging  from  — 10  to  +10.  The,  subjects  were  college 
students,  39  men  and  39  women.  Analysis  ,ql  responses 
showed  that  the  relative  ratings  given  to  each  act  by  men 
an3~women  were  very  similar,  i.e.,  the  same  acts  were 
cEosen  as  Best  and  worst  by  both  sexes.  There  was,  how- 
ever, a  tendency  for  women  to  assign  more  extreme  ratings 
both  to  good  and  to  bad  acts.  The  men  gave  more  medi- 
ocre or  indifferent  ratings,  while  the  women  employed 
more  frequently  the  values  towards  the  — 10  and  +10 
ends  of  the  scale.  Although  the  procedure  was  crude  and 
the  number  of  cases  small,  these  findings  are  suggestive. 
The  tendency  of  the  female  group  to  be  more  critical, 
both  favorably  and  unfavorably,  is  in  line  with  the  tradi- 
tional belief  that  women  are  the  custodians  of  the  social 
mores,  as  well  as  with  the  common  practice  of  attaching  a 
greater  ethical  or  moral  significance  to  the  activities  of 
women  than  to  those  of  men. 

In  an  extensive  series  of  tests  by^HartsljLQjcne, ,May,  and 
Shuttleworth  (16)  ,  on  approximately  85o_xkmejQitary 
schoolchildren  in  three  cities,  reliable  differences  in. favor 

olths gl4s,were,rfouBdiJl  ,mQ^l, Jk^Qwledga,, 

attitudes.  Several  tests  of  each  of  these  aspects  of  char- 
acter development  were  employed.  In  order  to  keep  as 
close  as  possible  to  the  children's  own  opinion,  the  tests 
were  worked  up  in  the  form  of  ballots  and  the  children 
were  asked  to  "vote"  on  each  item.  In  the  so-called 
"duties  test,"  for  example,  several  propositions  were 
given  with  the  request  that  the  subject  indicate  whether 
it  is  his  duty  to  do  these  things,  by  underlining  Yes,  No, 
or  S  (sometimes  yes  and  sometimes  no).  Some  of  the 
items  in  this  test  were  as  follows  (16,  pp.  46-47): 


1.  To  help  a  slow  or  dull  child  with  his  lessons  Yes        S        No 

2.  To  call  your  teacher's  attention  to  the  fact  if  you 

received  a  higher  grade  than  you  deserved  Yes        S        No 

3 .  To  smile  when  things  go  wrong  Yes       S       No 

4.  To  report  another  pupil  if  you  see  him  cheating       Yes        S        No 

In  total  scores  on  both  the  moral  judgment  and  social 
attitudes  tests,  the  differences  in  favor  of  the  girls  were 
4.31  times  as  large  as  their  standard  errors  and  can  there- 
fore be  regarded  as  perfectly  reliable.  The  investigators 
conclude  that:  "It  appears  on  the  surface  at  least  th#t 
girlFlireTnore  sensitive  to  both  conventional  and  ideal 
social  standards  than  boys"  (16,  p,  119). 

Certain  significant  sex  differences  have  alfco  been  dis- 
covered in  objective  behavioral  tests  of  character.  In  a 
series  of  investigations  by  Hartshorne  and  May  (14,  15), 
tests  were  devised  for  the  measurement  of  "deceit?  in- 
cluding cheating,  lying,  and  stealing,  "service"  including 
cooperative  and  charitable  behavior,  and  "self-control," 
including  persistence  and  inhibition.  Among  the  special 
advantages  of  these  tests  may  be  mentioned  the  fact  that 
the  subjects  did  not  realize  that  they  were  being  tested 
or  that  their  actions  could  be  detected.  All  observations, 
furthermore,  were  made  in  the  course  of  ordinary  every- 
.day  activities  of  the  children,  including  school  work,  home- 
work assignments,  athletics,  and  party  games.  Data  on  de- 
ceit were  collected  on  some  10,865  elementary  school  pupils 
in  several  parts  of  the  country.  For  the  main  studies  on 
service  and  self-control,  about  900  children  were  employed. 

No  consistent  sex  difference  in  deceptive  .behayigj  was 
4is£QYered.  Analysis*  of  "separate  tests  showed  that:  "On 
some  tests  the  girls  are  more  dishonest,  whereas  on  others 
the  boys  show  the  greater  tendency  to  deceive"  (14, 
p.  1 68) .  InjgiftjLjiAlys^. of  $ewfc&<wd~^^  com- 

were  made,bpt;h  in  test  scores 


among  classmates  and  teachers.  The  final  summary  data 
on  sex  differences  in  these  traits  are  shown  below  (cf.  15, 
pp.  156,  380,  382): 

Difference  n.       .        .  n._ 
*?_._              _  .          Direction  of  Difference 

S.E.  of  Difference 

Total  service  scores  1.9  Girls  more  cooperative 

Reputation  for  service  7.9  Girls  more  cooperative 

Total  persistence  scores  1.7  Girls  more  persistent 

Reputation  for  persistence  7.6  Girls  more  persistent 

Total  inhibition  score  5.5  Girls  better  inhibited 

Reputation  for  inhibition  5.0  Girls  better  inhibited 

It  will  be  noted  that  all  of  the  above  differences  are  in 
favor  of  the  girls.  In  service  and  persistence,  however, 
the  differences  in  total  score  are  not  reliable.  The  relative 
standing  of  the  two  sexes  also  varied  from  test  to  test. 
In  inhibition,  on  the  other  hand,  the  girls  were  reliably 
superior  in  total  score  and  consistently  superior  on  each 
individual  test.  The  more  successful  adjustment  of  girls 
to  the  school  situation  may  be  owing  partly  to  these  per- 
sonality differences.  It  is  also  interesting  to  note  in  this  con- 
nection that  in  reputation  the  girls  excel  the  boys  markedly 
in  all  traits.  This  too  may  influence  their  school  success. 

The  discrepancy  between  reputation  and  performance 
is  J^$^f  interest  In  relation  to  socialpressure.  Kiia^rbe 
that  with  increasing  age  the  cumulative  force  of  social- 
expectancy  becomes  more  effective  and  the  discrepancy 
between  behavior  and  traditional  belief  is  les&enedu  With  • 
this  would  come  an  increasing  differentiation  between  the 
sexes.  Until  similar  behavior  tests  are  made  on  adult 
subjects,  these  questions  cannot  be  answered. 


In  general  emotional  adjustment  and  nature  of  emo- 
tional response,  our  cultural  tradition  has  taught  us  to 


expect  pronounced  differences  between  man  and  woman. 
The  recent  development  of  standardized  personality  tests 
and  their  application  to  college  groups  have  revealed  a 
number  of  significant  sex  differences.  The  belief  in  the 
greater  emotional  instability  of  the  female  has  been  re- 
flected in  test  scores.  Thus  the  responses  of  about  600 
male  and  400  female  college  students  on  one  of  the  Col- 
gate Mental  Hygiene  Tests  (22)  indicated  an  average  of 
20%  more  psychasthenic  symptoms  among  women. 

In  an  investigation  on  introversion  (17),  100  college 
men  and  100  college  women  rated  themselves  and  were 
also  rated  individually  by  two  associates  on  54  typical 
introvert  traits.  No  significant  sex  difference  was  found 
in  introversion-extroversion,  the  average  scores  of  men 
and  women"  beihg^  It. 41  arid  11.12,  respectively.  But 
further  analysis  of  responses  showed  a  sex  difference  in 
another  trait  which  cut  across  introversion-extroversion 
ani&eemecl  to  bear  no  relation  to  it.  This  trait  is  concerned 
predominantly  with  social  relations.  The  introvert  traits 
marked  most  often  by  men  were  those  which  would  in,ter- 
ferve  with  social  adjustments.  Those  marked  most  often 
by  women,  on  the  other  hand,  were  such  as  to  interfere 
with  efficient  work.  A  few  examples  will  illustrate  this 

Typical  "masculine"  symptoms  of  introversion: 

1.  Outspoken 

2.  Works  things  out  on  own  hook;  hesitates  to  accept  help 

3.  Keeps  in  background  on  social  occasions 

4.  Conservative  and  painstaking  in  dress 

5.  Introspective 

Typical  "feminine"  symptoms  of  introversion: 

1.  Shrinks  when  facing  a  crisis 

2.  Works  by  fits  and  starts 

3.  Has  ups  and  downs  in  mood  without  apparent  cause 

4.  Feels  hurt  readily 

5.  Hesitates  in  making  decisions  on  ordinary  matters 


Comparisons  of  male  and  female  college  students  have 
also  been  made  on  the  Bernreuter  Personality  Inventory. 
This  test,  consisting  of  125  questions,  can  be  scored  with 
six  different  keys  for  as  many  different  traits.1  The 
average  scores  obtained  by  male  and  female  college  groups, 
as  well  as  the  critical  ratios  2  of  the  differences,  are  shown 
below.  The  number  of  cases  in  these  groups  varied  from 
144  to  658. 



Average  Score 
Male       Female 

S.E.  of  Difference 

Direction  of 

BiN:  Neuroti-      -57.3      -42.8 

Women  more 
neurotic,  or 

Men  more  self- 

Women  more 

Men  more 

Men  more  self- 

•Women  more 
and  socially 

In  all  but  .the  sociability  scale,  the  sex  differences  are 
perfectly  reliable.    This  trait  may  include  more  than  one 

1  The  fact  that  some  of  these  traits  have  recently  been  shown  to  be  mutually 
interdependent  introduces  unnecessary  duplication  in  the  scores  but  in  no  way 
vitiates  the  comparisons  made.  All  six  scores  are  'reported  here  for  the  sake  of 

2  Computed  by  the  writer  from  Bernreuter's  published  data — see  manual  of 
directions  and  norms,  October,  1934. 

B2S:  Self- 


Bsl :  Introver- 

B4D:  Dom- 

FiC:  Con- 

F2S:  -Socia- 

27-o  6.8 

-25.6  -14.7 

45-9  30.6 

-51-5  8.7 

-25-9  --3I-I 








counteracting  factor  in  which  the  sexes  might  differ  sig- 
nificantly. In  introversion,  it  will  be  noted  that  women 
obtain  significantly  higher  average  scores.  This  seems  to 
conflict  with  the  findings  reported  above.  The  explanation 
of  this  discrepancy  may  lie  in  the  particular  choice  of 
items  in  the  two  tests.  Thus  Bernreuter's  test  may  be 
overweighted  with  the  more  "feminine"  type  of  introvert 
items.  In  all  studies  on  personality,  it  is  difficult  to  gen- 
eralize regarding  trait  differences  because  of  the  specificity 
of  responses.  The  sex  differences  in  confidence  and  self- 
sufficiency  on  the  Bernreuter  test  are  particularly  large. 
The  differences  in  dominance  and  neuroticism  are  clearly 
reliable,  but  smaller.  It  should  be  kept  in  mind,  of  course, 
that  these  data  apply  only  to  college  students,  A  differed 
picture  might  be  presented  by  other  adult  groups.  It  is 
unlikely,  however,  that  the  sex  differences  in  non-college 
groups  should  be  smaller,  since  the  environment  of  college 
students  is  probably  more  uniform  than  that  of  other 
male  and  female  adults  in  our  society. 

Interesting  comparative  data  on  the  emotional  differ- 
ences between  boys  and  girls  at  successive  ages  have  been 
obtained  with  the  Woodworth-Mathews  questionnaire 
(28).  This  is  an  adaptation  of  the  Woodworth  psycho- 
neurotic  inventory  for  use  on  children.  It  consists  of 
100  questions  dealing  with  the  following  classes  of 
symptoms : 

1.  Fears,  worries,  perseverations 

2.  Physical  symptoms,  such  as  muscular  twitchings 

3.  Unhappiness,  anti-social  moods 

4.  Dreams,  phantasies,  sleep  disturbances 

This  test  was  administered  to  575  boys  and  558  girls  be- 
tween the  ages  of  9  and  19.  The  median  number  of  neu- 
rotic symptoms  reported  by  the  boys  was  20,  and  by  the 


girls  25.5.  This  sex  difference  is  in  agreement  with  the 
findings  on  adults.  But  an  analysis  of  scores  in  different 
age  groups  brought  out  certain  rather  suggestive  facts. 
The  changes  in  score  with  age  among  boys  and  girls  are 
shown  graphically  in  Figure  49. 

It  will  be  seen  that  at  age  10  the  median  number  of 
symptoms  reported  by  the  boys  is  greater  than  that  re- 




Z.  20 



10        11        12       13 


15       16       17       18 

28,  p.  21.) 

ported  by  the  girls.  At  u  there  is  no  appreciable  sex  dif- 
ference, and  beyond  this  the  girls  show  an  increase  in 
emotional  instability  and  the  boys  a  decrease.  This,  to  be 
sure,  is  what  would  be  expected  from  an  analysis  of  the 
environments  of  the  two  sexes.  The  emotional  problems 
and  affective  milieux  of  boys  and  girls  are  quite  similar  in 
early  childhood.  Gradually,  however,  the  environmental 
differentiation  becomes  pronounced  and  this  is  reflected 
in  the  increasing  divergence  of  the  sexes  in  emotional 


A  very  compr^^  to  the  problem  of 

differences  in  personality  traits  has  recently  been  made 

,(„„,.,,  ...  ***»-•  .1  **!,„.  .,,,.',  ,-^/f  !-,  ^  .«',/<.  ;  f  «.-<>  "<-V  <  ',,;,„,!  .,„      :  -  ,->,.„,.     ,      .',      <      '!.;,:*»/' 


by  Terman  and  Miles  (46).  The  aim  of  this  investigation 
was  to  devise  a  scale  which  would  differentiate  clearly 
between  the  characteristic  male  and  female  patterns  of 
response.  After  an  exhaustive  survey  of  the  literature 
and  prolonged  research,  items  were  selected  which  re- 
vealed the  most  pronounced  differences  between  repre- 
sentative samplings  of  the  two  sexes  in  our  society.  Data 
were  collected  on  many  hundreds  of  subjects,  including 
elementary,  high  school,  college,  and  graduate  students, 
unselected  adults,  members  of  several  occupations,  and 
specially  selected  groups  such  as  athletes,  juvenile  delin- 
quents, and  adult  homosexuals. 

Seven  tests  were  included  in  the  final  scale.  The  first 
is  a  word  association  test  and  the  second  an  ink-blot  asso- 
ciation test.  In  both  tests  the  response  is  of  the  multiple 
choice  form,  the  alternatives  under  each  item  including 
distinctly  masculine  as  well  as  distinctly  feminine  re- 
sponses. An  information  test,  drawing  upon  a  wide  variety 
of  fields,  is  likewise  of  the  multiple  choice  variety.  In  a 
test  of  emotional  and  ethical  response,  the  subject  is  re- 
quired to  check  the  degree  to  which  a  given  situation  pro- 
vokes in  him  emotions  of  anger,  fear,  disgust,  or  pity;  in 
the  second  part  of  this  test,  the  degree  of  moral  seriousness 
of  each  given  type  of  behavior  is  to  be  indicated  by  check- 
ing 3,  2,  I,  or  o.  An  interest  test  follows,  in  which  the  sub- 
ject expresses  his  liking,  disliking,  or  indifference  towards 
a  wide  range  of  items,  including  occupations,  people, 
games  and  amusements,  movies,  magazines,  school  studies, 
books  and  literary  characters,  travel  and  sightseeing  pref- 
erences, and  special  interests.  The  sixth  is  a  test  of  opin- 
ions^ in  which  a  series  of  common  beliefs  and  superstitions 
are  to  be  marked  true  or  false.  The  last  is  a  test  of  in- 
trovertive  response^  the  items  having  been  selected  from 
several  current  scales  of  introversion-extroversion.  A 


composite  weighted  score  on  all  seven  tests  can  be  ob- 
tained for  each  subject. 

This  scale  proved  very  successful  in  differentiating 
between  the  responses  of  male  and  female  groups.  All 
sex  differences  in  average  score  were  statistically  reliable, 
the  smallest  being  3.4  times  as  large  as  its  standard  error. 
An  intensive  analysis  of  the  characteristic  male  and  fe- 
male responses  on  each  test  revealed  a  distinct  picture  of 
the  "temperaments"  of  the  two  sexes  in  our  culture. 
Terman  and  Miles  summarize  these  differences  as  follows: 

From  whatever  angle  we  have  examined  them  the  males 
included  in  the  standardization  groups  evinced  a  distinctive 
interest  in  exploit  and  adventure,  in  outdoor  and  physically 
strenuous  occupations,  in  machinery  and  tools,  in  science, 
physical  phenomena,  and  inventions;  and,  from  rather  occa- 
sional evidence,  in  business  and  commerce.  On  the  other 
hand,  the  females  of  our  groups  have  evinced  a  distinctive 
interest  in  domestic  affairs  and  in  aesthetic  objects  and  oc- 
cupations; they  have  distinctly  preferred  more  sedentary 
and  indoor  occupations,  and  occupations  more  directly  min- 
istrative,  particularly  to  the  young,  the  helpless,  the  dis- 
tressed. Supporting  and  supplementing  these  are  the  more 
subjective  differences — those  in  emotional  disposition  and 
direction.  The  males  directly  or  indirectly  manifest  the  greater 
self-assertion  and  aggressiveness;  they  express  more  hardi- 
hood and  fearlessness,  and  more  roughness  of  manners,  lan- 
guage, and  sentiments.  The  females  express  themselves  as  more 
compassionate  and  sympathetic,  more  timid,  more  fastidious, 
and  aesthetically  sensitive,  more  emotional  in  general  (or 
at  least  more  expressive  of  the  four  emotions  considered), 
severer  moralists,  yet  admit  in  themselves  weaknesses  in 
emotional  control  and  (less  noticeably)  in  physique  (46, 
pp.  447-448). 

In  regard  to  the  origin  of  such  differences,  the  authors 
offer  only  tentative  hypotheses.      The  role  of  cultural 


factors  is,  however,  constantly  admitted  throughout  their 
discussion.  Several  lines  of  evidence  are  cited  which  sug- 
gest the  importance  of  nurture  in  the  development  of 
whatln^our  society  is  regarded  as  a  typically  "masculine" 
or  "feminine"  personality.  Thus  the  masculinity-fem- 
ininity index  was  found  to  be  associated  with  amount  of 
education,  occupation,  and  domestic  milieu.  Such 
factors  as  the  death  of  one  parent,  excessive  or  exclusive 
association  with  one  or  another  parent,  and  predominance 
of  brothers  or  sisters  among  the  siblings,  were  much  more 
closely  linked  with  masculinity-femininity  scores  than 
were  physical  traits.  Highly  intelligent  and  well-educated 
women  tended  to  score  more  " masculine"  than  their  sex 
norms.  Similarly,  " cultured"  men,  i.e.,  men  who  had 
cultivated  avocational  interests  of  an  artistic  or  intellec- 
tual sort,  tended  to  test  more  "feminine."  Thus  the 
equalizing  influence  of  specific  training  or  experience  in 
the  two  cases  seemed  to  bring  about  a  convergence  of 
the  temperamental  qualities  of  the  two  sexes. 


1.  Allen,  C.  N.    "Studies  in  Sex  Differences,"  Psychol  Bull, 
1927,  24,  294-304. 

2.  Allport,  G.  W.,  and  Vernon,  P.  E.    "A  Test  for  Personal 
Values,"  /.  Abn.  and  Soc.  PsychoL,  1931,  26,  231-248. 

3.  Batson,  W.  H.     "The  South  Dakota  Group  Intelligence 
Test  for  High  School/'  Sch.  and  Soc.,  1920,  15,  311-315. 

4.  Book,  W.  F.,  and  Meadows,  J.  L.   "Sex  Differences  in  5929 
High  School  Seniors  in  Ten  Psychological  Tests,"  /.  AppL 
Psyckol.,  1928,  12,  56-81. 

5.  Burt,  C.  Mental  and  Scholastic  Tests.  London:  King,   1922. 
Pp.  432. 

6.  Burt,  C.,  and  Moore,  R.  C.    "The  Mental  Differences  be- 
tween the  Sexes,"  /.  Exper.  Ped.^  1912,  I,  273-284;  355-388. 

7.  Cameron,  A.  E.   "  Comparative  Stud7  of  the  Mathematical 


Ability  of  Boys  and  Girls  in  the  Secondary  Schools,"  Brit. 
J.  Psychol.,  1925,  1 6,  29-49. 

8.  Commins,  W.  D.    "More  about  Sex  Differences,"  Sch.  and 
Soc.,  1928,  28,  599-600. 

9.  Fauth,    E.      "Testuntersuchungen   an   Schulkindern   nach 
Methode  des  fortlaufenden  Addierens,"  Arch.  f.d.  ges.  Psy- 
choL,  1925,  5 *,  I~20- 

10.  Gaw,  F.   "  A  Study  of  Performance  Tests,"  Brit.  J.  Psychol., 
1925,  *5,  374-392. 

11.  Gilbert,  J.  A.    "Researches  upon  School  Children  and  Col- 
lege Students,"  Univ.  Iowa  Stud,  in  Psychol.,  1897,  I.  Pp.  39. 

12.  Goodenough,  F.  L.    "The  Consistency  of  Sex  Differences  in 
Mental  Traits  at  Various  Ages,"  Psychol.  Rev..,  1927,  34, 

13.  .   "The  Kuhlman-Binet  Tests  for  Children  of  Preschool 

Age:  a  Critical  Study  and  Evaluation,"  Univ.  Minn.  Inst. 
Child  Welfare,  Man.  Series,  No.  2,  1927. 

14.  Hartshorne,  H.,  and  May,  M.  A.    Studies  in  the  Nature  of 
Character.     Vol.  I:  Studies  in  Deceit.     N.  Y.:  Macmillan, 
1928.   Pp.  414  +  306. 

15.  Hartshorne,  H.,  May,  M.  A.,  and  Mailer,  J.  B.    Studies  in 
the  Nature  of  Character.   Vol.  II :  Studies  in  Service  and  Self- 
Control.  N.  Y. :  Macmillan,  1929.  Pp.  559. 

16.  Hartshorne,  H.,  May,  M.  A.,   and   Shuttleworth,   F.   K. 
Studies  in  the  Nature  of  Character.    Vol.  Ill:  Studies  in  the 
Organization  of  Character.  N.  Y. :  Macmillan,  1930.  Pp.  503. 

17.  Heidbreder,  E.   "Introversion  and  Extroversion  in  Men  and 
Women,"  /.  Abn.  and  Soc.  Psychol.,  1927,  22,  52-61. 

18.  Hollingworth,  L.  S.     "Sex  Differences  in  Mental  Traits," 
Psychol.  Bull,  1916,  13,  377-384. 

ICj.  .     "Comparison  of  the  Sexes  in  Mental  Traits,"  Psy- 
chol. Bull,  1918,  15,  427-432. 

20.  Hunt,    T.     "The    Measurement    of    Social    Intelligence," 
/.  Appl.  Psychol.,  1928,  12,  3i7~334- 

21.  Kuper,  G.  M.    "Group  Differences  in  the  Interests  of  Chil- 
dren," /.  Phil.,  Psychol.,  and  ScienL  Method,  1912,  9,  376- 


22.  Laird,  D.  A.,  and  McClumpha,  T.     "Sex  Differences  in 
Emotional  Outlets,"  Science,  1925,  62,  292. 

23.  Landis,  C     "National  Differences  in  Conversations/9  /. 
Abn.  andSoc.  PsychoL,  1927,  21,  354-357- 

24.  Landis,  H.  M.,  and  Burtt,  H.  E.    "A  Study  of  Conversa- 
tions/' /.  Comp.  Psychol.,  1924,  4,  81-89. 

25.  Lentz,  T.   F.      "Sex  Differences   in  School  Marks  with 
Achievement  Test  Scores  Constant,"  Sch.  and  Soc.,  1929, 
29,  65-68. 

26.  Lincoln,  E.  A.     Sex  Differences  in  the  Growth  of  American 
School  Children.   Baltimore:  Warwick  and  York,  1927.   Pp. 

27.  Loutitt,  C.  M.     "A  Bibliography  of  Sex  Differences  in 
Mental  Traits/'  TV.  Sch.  Bull.,  1925,  22,  129-138. 

28.  Mathews,  E.     "A  Study  of  Emotional  Stability  in  Chil- 
dren/' /.  Delinq.,  1923,  8,  1-40. 

29.  McCarthy,  D.    "Language  Development."   Ch.  8  in  Hand- 
book of  Child  Psychology.    C.  Murchison,  ed.    Worcester, 
Mass.:  Clark  Univ.  Press,  1933.   Pp.  956. 

30.  McFarlane,  M.    "A  Study  of  Practical  Ability,"  Brit.  J. 
PsychoL,  Mon.  Suppl.,  1925,  3.  Pp.  75. 

31.  Miles,  C.  C.  "Sex  in  Social  Psychology."  Ch.  16  in  Handbook 
of  Social  Psychology.   C.  Murchison,  ed.  Worcester,  Mass.: 
Clark  Univ.  Press,  1935.    Pp.  1195. 

32.  Miles,  C.  C.,  and  Terman,  L.  M.    "Sex  Differences  in  the 
Association  of  Ideas,"  Amer.  J.  Psychol.,  1929,  41,  165-206. 

33.  Moore,  H.  T.   "Further  Data  concerning  Sex  Differences," 
J.  Abn.  and  Soc.  Psychol..,  1922,  17,  210-214. 

34.  Mulhall,  E.  F.  "Tests  of  the  Memories  of  School  Children/5 
/.  Educ.  PsychoL,  1917,  8,  295-302. 

35.  N.  Y.  College  Entrance  Examination  Board:  Reports  of  the 
Commission  on  Scholastic  Aptitude  Tests ,  1930. 

36.  Paterson,  D.  G.,  and  Langlie,  T.  A.   "The  Influence  of  Sex 
on  Scholarship  Ratings/'  Educ.  Admin,  and  Super.)  1926, 
12,  458-468. 

37.  Paterson,  D.  G.,  et  al    Minnesota  Mechanical  Ability  Tests. 
Minneapolis:  Univ.  Minn.  Press,  1930,   Pp.  586, 


38.  Porteus,  S.  D.     "The  Measurement  of  Intelligence:  643 
Children    Examined    by   the    Binet    and   Porteus   Tests," 
/.  Educ.  PsychoL,  1918,  9,  13-31. 

39.  Pressey,  L.  W.     "Sex  Differences  Shown  by  2544  School 
Children  on  a  Group  Scale  of  Intelligence,  with  Special  Refer- 
ence to  Variability,"  /.  AppL  PsychoL,  1918,  2,  323-340. 

40.  Pyle,  W.  H.    The  Examination  of  School  Children:  a  Manual 
of  Directions  and  Norms.  N.  Y.:  Macmillan,  1913.  Pp.  70. 

41.  Seashore,  C.  E.    "The  Measurement  of  Pitch  Discrimina- 
tion," PsychoL  Mon.y  1910,  13,  No.  i,  21-59. 

42.  Snyder,  A.,  and  Dunlap,  K.   "A  Study  of  Moral  Evaluation 
by  Male  and  Female  Students,"  /.  Comp.  PsychoL,  1924, 
4,  289-324. 

43.  Stenquist,   J.      "Measurements   of  Mechanical   Ability," 
Teachers  College,  Columbia  Univ.,  Contrib.  to  Educ.,  No. 
130,  1923.  Pp.  101. 

44.  Terman,  L.  M.     The  Measurement  of  Intelligence.    N.  Y. : 
Houghton  Mifflin,  1916.  Pp.  362. 

45.  Terman,  L.  M.,  and  Cuneo,  I.  " Stanford-Binet  Tests  of  1 12 
Kindergarten   Children  and  77  Repeated   Retests,"   Ped. 
Sem.,  1918,  25,  414-428. 

46.  Terman,  L.  M.,  and  Miles,  C.  C.     Sex  and  Personality: 
Studies  in  Masculinity  and  Femininity.    N.  Y.:  McGraw- 
Hill,  1936.  Pp.  600. 

47.  Terman,  L.  M.,  et  al.    The  Stanford  Revision  and  Extension 
of  the  Binet-Simon  Scale  for  Measuring  Intelligence.    Balti- 
more: Warwick  and  York,  1917.  Pp.  179. 

48.  Terman,  L.  M.,  et  al.     Genetic  Studies  of  Genius.  Vol.  I. 
Stanford  Univ.,  Calif.:  Stanford  Univ.  Press,  1925.  Pp.  648. 

49.  Thompson,  H.  B.   The  Mental  Traits  of  Sex.  Chicago:  Univ. 
Chicago  Press,  1903.  Pp.  188. 

50.  Thorndike,  E.  L.,  et  al.     "Sex  Differences  in  Status  and 
Gain  in  Intelligence  Test  Scores  from  13-18,"  Ped.  Sem., 
1926,  33,  167-181. 

51.  Webb,  P.  E.    "A  Study  of  Geometric  Abilities  among  Boys 
and  Girls  of  Equal  Mental  Abilities,"  /.  Educ.  Res.,  1927, 
15,  256-262. 


52.  Wellman,  B.  L.    "Sex  Differences."   Ch.  15  in  Handbook  of 
Child  Psychology.    C.  Murchison,  ed.     Worcester,  Mass.: 
Clark  Univ.  Press,  1933.   Pp.  956. 

53.  Whipple,  G.  M.     Manual  of  Physical  and  Mental  Tests. 
Baltimore:  Warwick  and  York,  1921.  Parts  I  and  II. 

54.  .    "Sex  Differences  in  Intelligence  Test  Scores  in  the 

Elementary  School,"  /.  Educ.  Res.,  1927,  15,  111-117. 

ijcj%  f      "Sex  Differences   in  Army  Alpha   Scores   in   the 

Secondary  School,"  /.  Educ.  Res.,  1927,  15,  269-275. 

56.  Wissler,   C.      "The   Correlation  of  Mental   and  Physical 
Tests,"  Psychol  Mon.,  1901,  3.  Pp.  62. 

57.  Woolley,  H.  T.    "Psychology  of  Sex,"  Psychol.  Bull,  1910, 
7,  33S-342. 

58.  .    "Psychology  of  Sex,"  PsychoL  £ull.9  1914,  n,  353- 




The  comparative  evaluation  of  the  races  of  man  has 
long  been  a  subject  of  keen  interest  and  lively  controversy. 
It  is  an  interesting  commentary  upon  human  thought 
that  nearly  all  theories  of  racial  inequality  proclaim  the 
superiority  of  the  particular  race  of  their  respective  ex- 
ponents. Thus  Aristotle  (cf.  I,  pp.  318-320)  endeavored 
to  demonstrate  that  the  intellectual  leadership  of  the 
Greeks  must  of  necessity  follow  from  their  favorable 
geographical  location.  He  argued  that  the  peoples  in- 
habiting the  colder  regions  of  northern  Europe,  although 
outstanding  for  bravery  and  physical  prowess,  were  in- 
tellectually incapable  of  a  high  degree  of  political  organiza- 
tion or  leadership.  Similarly,  the  Asiatics,  although  in- 
tellectually keen  and  inventive, 'lacked  spirit.  The  Greeks 
alone,  being  geographically  intermediate,  were  endowed 
with  the  proper  balance  of  these  traits  and  were  thus  by 
nature  fitted  to  rule  the  earth.  Similar  claims  have  been 
made  for  such  groups  as  the  Arabians,  the  Romans,  the 
French,  the  Anglo-Saxon,  the  "White"  race  as  distin- 
guished from  those  having  a  different  skin  pigmentation, 
the  Nordics,  the  Alpines,  the  Mediterraneans,  and  various 

Outstanding  among  such  theories,  because  of  its  wide- 
spread popularization,  is  that  proposed  by  de  Gobineau 
(10)  in  the  nineteenth  century  and  subsequently  expanded 
by  Chamberlain  (7).  This  doctrine  had  numerous  followers 
who  reformulated  it  and  developed  it  along  various  lines. 


Its  basic  contention,  however,  is  the  superiority  of 
the  Nordic  or  "Aryan"  race,  a  loosely  and  ambiguously 
defined  group  whose  descendants  are  now  supposed  to 
inhabit  for  the  most  part  the  countries  of  northern 
Europe.  The  array  of  evidence  cited  in  support  of  this 
theory  is  incomplete  and  one-sided  at  its  best  and  fantastic 
and  mythical  at  its  worst.  The  concepts  involved  in 
such  a  theory  will  be  critically  examined  in  the  course  of 
the  subsequent  discussion. 

At  present,  race  problems  are  a  particularly  lively  issue. 
Outworn  and  forgotten  theories  are  being  revived  in  the 
attempt  to  rationalize  political  actions  and  policies.  Under 
the  stress  of  emotional  appeal,  it  is  especially  difficult 
to  carry  on  unbiased  and  objective  analysis  of  facts.  It 
is  one  of  the  earmarks  of  prejudice  to  draw  logically  un- 
warranted inferences  from  the  data  at  hand.  In  a  current 
test  (34)  for  the  measurement  of  the  prejudice-fairminded- 
ness  variable,  for  example,  one  of  the  subtests  is  based  on 
just  such  behavior.  The  subject  is  given  certain  facts 
bearing  upon  controversial  issues,  with  the  instructions 
to  check  any  of  the  proposed  conclusions  which  seem  to 
him  to  follow  directly  from  the  given  data,  regardless  of 
their  truth  or  falsity  in  general.  The  individual  who  is 
biased  or  who  responds  emotionally  to  any  of  the  issues 
involved  will  ignore  the  limitations  of  the  facts  actually 
presented  and  will  generalize  far  beyond  them.  The  pro- 
cedure in  this  test  presents  a  close  parallel  to  what  proba- 
bly occurs  all  too  often  in  the  interpretation  of  data  on 
such  emotionally  toned  issues  as  race  differences. 

Under  such  conditions  it  is  essential  to  emphasize  and 
to  bear  clearly  in  mind  the  possible  vitiating  factors  and 
sources  of  error  in  the  data.  As  in  all  group  comparisons, 
studies  on  race  differences  must  take  into  account  selective 
factors  and  adequacy  of  sampling,  overlapping  of  dis- 


tributions,  reliability  of  an  obtained  difference,  inaccuracy 
or  ambiguity  of  the  measuring  instrument,  and  other 
similar  factors  which  have  already  been  discussed  and 
illustrated  (cf.  Ch.  XIV).  It  is  probably  not  an  exaggera- 
tion to  state  that  failure  to  consider  such  factors  has  in- 
validated the  large  majority  of  investigations  which  pur- 
port to  have  established  a  racial  difference  in  one  or 
another  behavioral  trait. 

Racial  comparisons  are  an  extremely  difficult  problem 
of  differential  psychology.  In  addition  to  the  above- 
mentioned  sources  of  error  which  they  share  with  all 
group  comparisons,  studies  on  race  differences  are  handi- 
capped by  special  difficulties  inherent  in  every  phase  of 
the  problem.  Thus  it  has  proved  a  difficult  matter  in 
such  studies  to  decide  whom  to  measure,  what  to  measure, 
and  how  to  measure  it.  These  difficulties  will  be  analyzed 
in  the  present  and  subsequent  chapter.  The  first  of  these 
two  chapters  will  be  concerned  with  questions  of  whom 
to  measure,  or  the  selection  and  classification  of  subjects 
in  racial  studies.  In  the  second  chapter  will  be  discussed 
some  of  the  major  problems  which  arise  in  the  efforts  to 
measure  and  compare  widely  diverse  groups. 

The  data  of  investigations  on  race  differences  have  been 
grouped  about  these  methodological  questions.  No  gen- 
eral summary  of  findings  or  intellectual  hierarchy  of  racial 
groups  is  presented  because,  although  apparently  useful 
as  a  mnemonic  device,  such  a  tabulation  would  be  highly 
misleading  and  of  very  dubious  value.  Isolated  facts  are 
particularly  misleading  in  racial  comparisons  and  should 
at  all  times  be  evaluated  in  terms  of  the  conditions  under 
which  they  were  collected.  Conclusions  on  race  differences 
will  therefore  be  drawn  only  in  the  light  of  a  critical  analysis 
of  the  entire  problem  and  will  not  be  divorced  from  their 
limiting  conditions.  No  attempt  has  been  made,  further- 


more,  to  survey  the  vast  array  of  investigations  on  psy- 
chological differences  among  racial  groups.  For  summaries 
and  more  extensive  discussions  of  this  problem,  the  reader 
is  referred  to  a  review  (14)  and  book  (16)  by  Garth  and 
to  the  more  recent  and  very  stimulating  book  by  Klineberg 
(24).  For  an  orientation  into  the  general  problem  of  race, 
books  on  anthropology,  such  as  those  of  Boas  (4)  and 
Kroeber  (26),  may  be  consulted. 


Tradition,  prejudice,  and  the  snap  judgments  of  every- 
day observation  have  contributed  to  the  development  of 
a  concept  of  race  as  a  clearly  differentiated  and  easily 
identifiable  group,  possessing  distinctive  physical,  mental, 
and  temperamental  characteristics.  The  observations  of 
biologists,  anthropologists,  and  psychologists,  however, 
fail  to  support  such  a  view.1  The  classification  into  racial 
groups  is  essentially  a  biological  one  and  corresponds  to 
such  divisions  as  breed,  stock,  and  strain  in  infrahuman 
organisms.  In  its  simplest  terms,  any  definition  of  race 
implies,  a  certain  .community  of  physical  characteristics  based 
pnm^nly  up&ik  $  common  heredity,, 

The  task  of  race  classification  is  far  more  complex  than 
would  appear  from  the  glibness  with  which  individuals 
are  commonly  assigned  to  one  group  or  another.  The 
five-fold  classification  of  races  memorized  by  every  school- 
child  is  of  historical  interest  only.  This  division  can  be 
traced  to  Linnseus  (27),  the  great  classifier,  who  recog- 
nized four  races  of  men — Europeans  albus  (white),  Ameri- 
canus  rubescens  (red),  Asiaticus  fuscus  (yellow),  and 
Africans*  niger  (black).  A  fifth  group,  the  brown  race, 
was  subsequently  added  by  Blumenbach  (2),  who  also 

1  For  a  very  readable  account  of  many  of  the  difficulties  of  race  classification, 
see  Huxley  and  Haddon  (21). 


altered  the  terminology,  proposing  the  now  familiar  classi- 
fication into  Caucasian,  Mongolian,  American,  Ethiopian, 
and  Malayan.  This  classification  is  crude  and  superficial, 
as  will  shortly  become  apparent. 

The  essential  problem  in  the  classification  of  racial 
groups  consists  in  the  identification,  of  inheritable  physical 
characteristics  which  differ  clearly  from  one  group  to 
,  aootKer  End  which  may  thus  serve  as  criteria^  jace.  A 
wide  variety  of  such  criteria  have  been  proposed  and 
applied  (cf.  e.g.,  24,  26).  Skin  color  is  popularly  .employed 
as  one  of  the  most  obvious  means  of  racial  identification. 
In  spite  of  its  widespread  use,  however,  it  has  proved 
to  be  one  of  the  least  satisfactory  of  the  possible  criteria. 
It  is  a  well-established  fact  that  the  same  pigments  are 
present  in  all  human  skins  and  that  different  skin  colors 
result  from  varying  relative  amounts  of  each  pigment. 
For  this  reason,  there  is  found  a  complete  series  of  transi- 
tion shades,  making wexact  classification  very  difficult. 
Such  a  classification  is  also  rendered  somewhat  unstable 
by  the  fact  that  environmental  conditions,  such  as  ex- 
posure to  the  sun's  rays,  have  a  marked  effect  upon  skin 

Pigmentation,  of  <tk$ <eyes  has  proved ;a  somewhat  more 
fruitful  index,  insofar  as  it  is  unquestionably  an  hereditary 
tjrait.  In  the  same  connection  may  be  mentioaed.JwV 
color.  These  traits,  however,  are  also  difficult  to  describe 
quantitatively  because  of  continuous  gradations.  A 
further  difficulty  in  the  use  of  such  criteria  is  their  rela- 
tively narrow  distribution,  black  hair  and  eyes  being  the 
universal  rule  outside  of  a  single  stock. 

In  addition  to  coloring,  other  characteristics  of  the 
hair  have  been  employed  as  differentiating  signs.  The 
texture  of  the  hair  is  generally  regarded  as  a  valuable 
criterion  of  race  classification.  Fullness  of  the  beard  and 


degree  of  development  of  the  downy  hair  which  covers 
the  body  as  a  whole  are  also  coming  to  be  recognized  as 
useful  indices. 

Racial  groups  have  been  differentiated  on  the  basis  of 
gross  bodily  dimensions,  chief  among  which  is  stature. 
Group  differences  in  this  respect  are,  however,  surpris- 
ingly small  and  consequently  of  doubtful  value  in  racial 
identification.  Facial  and  cranial  measurements  have 
been  employed  to  somewhat  better  advantage.  Among 
the  former,  the  most  common  are  nasal  index,  which  ex- 
presses the  relative  length  and  breadth  of  nose,  and 
various  indices  of  prognathism,  or  the  degree  of  protru- 
sion of  the  jaws.  Cranial  capacity,  or  volume  of  the  skull, 
yields  rather  ambiguous  results  because  of  its  depend- 
ence on  general  body  size  and  because  of  the  wide  variation 
within  groups  with  consequent  overlapping  between  groups. 
Cephalic  index,1  on  the  other  hand,  has  proved  to  be 
one  of  the  most  satisfactory  criteria  of  classification  and 
is  now  widely  employed. 

In  view  of  the  relative  paucity  of  satisfactory  anatomical 
criteria,  attempts  have  been  made  to  evolve  physiological 
or  biochemical  schemas  of  classification.  It  has  been 
suggested,  for  example,  that  races  might  be  classified  on 
the  basis  of  the  ,  blood  groups  which  have  become 
familiar  in  connection  with  blood  transfusions  (cf.  19). 
These  blood  groupings  refer  to  the  agglutinative  reactions 
of  the  red  blood  corpuscles,  i-.e.,  the  tendency  of  such 
corpuscles  to  clump  together- when  the  blood  of  certain 
individuals  is  mixed  with  that  of  certain  other  individuals. 
It  is  now  well  known  that  there  exist  so-called  "universal 
donors"  whose  blood  can  be  safely  added  to  that  of  any 
individual,  as  well  as  "universal  recipients"  who  can 

,  „    ,    t.    .  ioo  X  head  width 

1  Cephalic  index  = head  1  n  th F°r  fuller  descriPtion?  see  Ch- 


safely  receive  the  blood  of  any  other  person.  There  are, 
In  addition,  classes  of  persons  who  can  receive  blood  from 
certain  specific  groups  only,  and  who  can  likewise  donate 
only  to  certain  groups.  Some  data  on  the  relative  inci- 
dence of  the  various  blood  groups  in  different  racial  stocks 
suggest  a  possible  relationship  between  the  two,  although 
tirejdiff^rentiation  is  by  no  means  clear.  A  serious  ob- 
jection to  such  a  classification  is  that  it  cuts  across  and 
conflicts  with  the  usual  anthropological  divisions  obtained 
with  other  criteria.  On  the  basis  of  blood  groupings,  very 
diverse  groups  would  be  lumped  together  indiscriminately. 

The  endocrine  glands  have  also  played  their  part  in  race 
classification.  Likenesses  have  been  noted,  for  instance, 
between  the  physical  and  alleged  mental  characteristics 
of  certain  groups  and  those  of  pathological  conditions 
associated  with  glandular  dysfunction  (cf.  e.g.,  22).  Thus 
a  parallel  has  been  drawn  between  the  cretin  and  the 
African  pygmy.  Pituitary  enlargement  has  been  attributed 
to  the  Hottentots  and  adrenal  deficiency  to  the  Negro. 
The  "childlike"  appearance  of  the  Chinese  has  been 
ascribed  to  an  over-active  thymus.  Such  methods  of 
classification  are  especially  questionable  for  two  reasons: 
they  take  a  superficial  and  partial  resemblance  as  their 
point  of  departure;  and  they  reason  from  pathological 
conditions  existing  within  a  single  group  to  the  normal 
characteristics  of  entire  groups. 

Finally,  mention  should  be  made  of  the  efforts  to  deal 
with  race^Jn^jter^^  (cf.  Ch.  IX). 

Kretschmer  (25),  for  example,  believes  the  ratio  of  lg£to- 
sojnes  and  pyknic§  to  differ  in  various  racial  groups  and 
offers  this  as  a  possible  explanation  of  the  psychological- 
differences  between  such  groups.  Others,  both  prior  and 
subsequent  to  Kretschmer,  have  attempted  similar  classi- 
fications. The  reader  should  recall  in  this  connection  the 


difficulty  of  finding  "pure  types"  and  the  absence  of  valid 
evidence  for  a  correlation  between  the  physical  charac- 
teristics of  such  types  and  any  of  their  alleged  mental 


In  addition  to  the  special  deficiencies  of  individual 
methods  of  classification  discussed  in  the  preceding  section, 
certain  major  difficulties  are  encountered  in  the  applica- 
tion of  all,  or  nearly  all,  criteria  of  race.  In  the  first  place, 
a  wide  variability  exists  within  any  one  group  in  respect 
to  any  trait.  Closely  related  to  this  is  the  marked  over- 
lapping between  different  groups  in  any  of  the  criteria 
msgtioned.  Thus,  although  two  groups  may  differ  sig- 
nificantly in  average  height,  individuals  can  readily  be 
found  in  the  " shorter  group"  who  are  taller  than  certain 
individuals  in  the  "taller  group."  This  obviously  makes 
group  delineations  indistinct  and  relatively  arbitrary. 

A  third  difficulty  is  the  inconsistency  frequently  found 
whetrmoi^tha^  one  criterion  is  employed!  An  Individual 
might  have  the  coloring  of  a  Nordic,  the  cephalic  index 
of  an  Alpine,  and  the  stature  of  a  Mediterranean.  Or  very 
dark  skin  pigmentation  and  woolly  hair  might  be  found 
in  association  with  Caucasian  features.  Such  instances 
are  frequent  and  cannot  be  dismissed  as  exceptions. 

Finally,  it  should  be  noted  that  many  of  the, .ajjeged 

raaalj:£.?iactenst}p8 yyiiich ^ere  formerly  heUsvacLto  be 

stable  and  innate  are  bqing  found ( .s^Qeptible  tQ  mviron- 
mmtal  "influences.  Even  such  apparently  "hereditary" 
traits  as  body  height,  skull  shape,  and  facial  conformation 
have  proved  to  be  dependent  in  part  upon  stimulating 
conditions  in  early  childhood.  This  was  clearly  demon- 
strated in  the  investigations  of  Boas  (3)  on  the  American- 
born  children  of  immigrants  from  several  European 


countries.  These  children  were  compared  with  foreign- 
born  children  from  the  same  countries,  who  were  also 
living  in  America.  Differences  were  found  in  stature, 
weight,  and  length  and  width  of  head. 

The  most  striking  indication  of  environmental  influence, 
however,  was  furnished  by  an  examination  of  the  cephalic 
indices  of  two  immigrant  groups  which  differ  markedly  in 
head  shape  (3).  American-born  and  foreign-born  boys 
were  compared  within  the  east  European  Hebrew  and  the 
Sicilian  groups  living  in  New  York  City.  The  former 
are  characteristically  round-headed,  having  a  high  cephalic 
index;  the  latter  are  characteristically  long-headed.  As 
will  be  seen  in  Table  XX,  residence  in  the  new  envirpn- 
itiDeftfc- tends'  to  make  the  Jewish  group  more  long-headed 
and  the"  Sicilian"  more  round-headed,  both  groups  con- 
verging toward  the  American  norna.  It  will  also  be  noted 

•~-<««»«,,S^.^w-S»-»«»v,«,««. ...— -  -•"*'  " 

that  those  boys  born  after  a  relatively  long  period  of 
American  residence  of  the  mother  exhibit  a  greater  change 
than  those  born  after  a  shorter  residence  period.  This 
was  also  found  to  be  the  case  in  the  data  on  other  immi- 
grant groups. 

That  these  physical  changes  were  the  result  of  changing 
environmental  conditions  rather  than  selective  factors 
was  clearly  demonstrated.  A  comparison  of  foreign-born 
persons  who  had  immigrated  at  different  periods  showed 
no  significant  differences  in  the  traits  under  consideration. 
The  measurement  of  American-born  and  foreign-born 
children  of  the  same  parents^  furthermore,  revealed  differ- 
ences in  the  expected  direction. 

The  results  of  Boas  have  subsequently  been  corrob- 
orated by  Guthe  (17),  who  compared  the  cephalic  indices 
of  187  Russian-born  Jewish  children  and  127  American- 
born  Jewish  children  in  Boston.  The  cephalic  indices 
found  by  Hirsch  (18)  on  American-born  children  of